Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
Arguments and amendments filed 9/3/2025 have been examined.
Claims 1 and 3 are amended.
Claim 6 is cancelled.
In this Office Action, claims 1-5, 7-8 are currently pending.
This Office Action is Final.
Response to Arguments
Applicant’s arguments with respect to claim(s) and the prior are rejection under 35 USC 103 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Applicant's arguments filed concerning rejections under 35 USC 101 have been fully considered but they are not persuasive.
As to the argument:
“Applicant respectfully contends that amended claim 1 does not fall into a judicial exception,
such as an abstract idea because the features recited in amended claim 1 cannot practically be certain methods of organizing human activity. Hence, Applicant respectfully submits that amended claim 1 as a whole is not directed to an abstract idea. See Step 2A, Prong One analysis stated in MPEP § 2164.04(a)(l). Further, even if amended claim 1 is considered to be directed to an abstract idea, Applicant respectfully contends that amended claim 1 integrates the alleged abstract idea into a practical application by demonstrating a technical improvement in an existing technology, such as improving the accuracy of the data extraction process. In particular, amended claim 1 specifically recites, inter alia, that the operation of automatically adding information to the database, and the operations of evaluation. Applicant respectfully submits that these features demonstrate a technical improvement, such as improving recognition rate and the accuracy of the data extraction process, as described in, for example, paragraph [70] and [78] of the original specification.”;
The Examiner respectfully disagrees.
As to Applicants assertion above, “In particular, amended claim 1 specifically recites, inter alia, that the operation of automatically adding information to the database, and the operations of evaluation”; however, here, nowhere does Applicant specially explain or claim the
algorithm or technique that allows the system to perform “automatically adding information”, note specifically the related rejection below under 35 USC 112, concerning the repeated use of “and/or” language in claim 1 (see “performing selecting only composition and/or property pairs”); which renders the claimed steps indefinite, and thus similarly unspecific for the purposes of analysis under 35 USC 101 abstract idea analysis; nor does Applicant explain or claim what algorithm performs the “automatically adding information” (see for example “performing selecting only composition and/or property pairs”). This blanket/generic claiming of
operations (where the algorithm mentioned in the claim is not claimed, identified, or explained")
is analogous to case law where the claims have been found to be abstract. Please see
Clarilogic v. Freeform Holdings (Court of Appeals for the Federal Circuit 2016-1781,
decided March 15, 2017), which recites at page 7:
"Peculiar to this case is that the algorithm engine mentioned in the claim is not
claimed, identified, or explained. To be sure, claiming an algorithm does not
alone render subject matter patent eligible. See Gottschalk v. Benson, 409 U.S.
63, 71-72 (1972). But a method for collection, analysis, and generation of
information reports, where the claims are not limited to how the collected
information is analyzed or reformed, is the height of abstraction.".
Similarly to the claims in Clarilogic, instant claim 1 is not limited to how the collected information is used for “automatically adding information”; or what algorithm performs the “automatically adding information” using collected information. Thus, to quote the court in Clarilogic,
"where the claims are not limited to how the collected information is analyzed or
reformed, is the height of abstraction".
Thus this argument is moot, the examiner remains unconvinced and the rejection under 35 USC 101 remains.
As to the argument that:
“Accordingly, Applicant respectfully contends that amended claim 1 specifically and practically sets forth the features that demonstrate how the alleged abstract idea is applied in a way that provides concrete benefits or solves specific problems in the relevant field. Hence, Applicant respectfully submits that amended claim 1 as a whole integrates the alleged abstract idea into a practical application. See Step 2A, Prong Two analysis stated in MPEP § 2106.04(d) Additionally, Applicant respectfully contends that, as discussed above, amended claim 1 provides a technical improvement in the relevant field and that the specification as filed identifies a technical problem and explains the details of an unconventional technical solution expressed in the claim or identifies technical improvements realized by the claim over the prior art, thereby adding inventive concepts to the alleged abstract idea. Therefore, Applicant respectfully submits that amended claim 1 recites significantly more than a judicial exception. See Step 2B analysis stated in MPEP § 2106.05(a).
For at least the above reasons, Applicant respectfully submits that amended independent claim 1 is subject matter eligible and that dependent claims 2-5, 7 and 8 are subject matter eligible at least by virtue of their dependence on amended independent claim 1. Thus, Applicant submits that the rejection of claims 1-5, 7 and 8 under 35 U.S.C. § 101 is overcome. The cancelation of claim 6 renders the rejection of claim 6 moot. Accordingly, Applicant respectfully requests reconsideration and withdrawal of the rejection under 35 U.S.C. § 101.”
The Examiner respectfully disagrees.
As to Applicant’s assertion above “amended claim 1 provides a technical improvement in the relevant field and that the specification as filed identifies a technical problem and explains the details of an unconventional technical solution expressed in the claim or identifies technical improvements realized by the claim over the prior art, thereby adding inventive concepts to the alleged abstract idea”; nowhere does Applicant specifically explain or connect the above assertions to the claim language, i.e. as to the assertions:
“amended claim 1 provides a technical improvement in the relevant field” ;
nowhere does Applicant explain what the technical improvement is, nor what claim limitations are related to the technical improvement; nor what specific “relevant field” is;
“and that the specification as filed identifies a technical problem and explains the details of an unconventional technical solution expressed in the claim”; again nowhere does Applicant explain or identify the “unconventional technical solution expressed in the claim” nor does Applicant identify or explain what specific portions/limitations of “the specification as filed identifies a technical problem and explains the details”; thus this argument is moot, the examiner remains unconvinced and the rejection under 35 USC 101 remains.
Priority
Acknowledgment is made of applicant's claim for foreign priority based on an application filed:
KR10-2023-0110403 REPUBLIC OF KOREA Filing Date 08/23/2023;
and
KR10-2024-0019940 REPUBLIC OF KOREA Filing Date 02/08/2024;
It is noted Applicant has filed two “Interim Copy of the Foreign Priority Document” on 9/3/2025;
however, that applicant has still not filed a certified copy of the above application(s) as required by 37 CFR 1.55.
Additionally, please note the two “PRIORITY DOCUMENT EXCHANGE FAILURE STATUS REPORT”(s), both dated 1/23/2025, noting:
“An attempt by the Office to electronically retrieve, under the priority document exchange program, the foreign application 10-2023-0110403 to which priority is claimed has FAILED on 01/23/2025.”;
and
“An attempt by the Office to electronically retrieve, under the priority document exchange program, the foreign application 10-2024-0019940 to which priority is claimed has FAILED on 01/23/2025.”; and
“Useful information is provided at the Electronic Priority Document Exchange (PDX) Program Website (https://www.uspto.gov/patents/basics/international-protection/electronic-priority-document-exchange-pdx), including practice tips for priority document exchange
(https://www.uspto.gov/patents/basics/international-protection/electronic-priority-document-exchange-pdx#Practice%20tips).
For further questions or assistance, please contact the Patent Electronic Business Center (EBC):
EBC Customer Support Center 1-866-217-9197 (toll-free)
571-272-4100 (local) M-F 6AM - Midnight (Eastern Time)
PDX@uspto.gov( email)”
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-5 and 7-8 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding claim 1, the phrase(s) “and/or” in the following limitations:
"selecting only composition and/or property pairs"; (Claim 1, lines 28-29)
“from a composition table and/or a property table,” (Claim 1, Lines 29-30)
“adding the selected composition and/or property pairs”; (Claim 1 line 30)
Render(s) the claim indefinite because it is unclear whether the limitation(s) following the phrase are part of the claimed invention. See MPEP § 2173.05(d).
The dependent claims inherit and do not correct the above defects, and thus are also rejected under 35 USC 112
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-5, 7-8 are rejected under 35 U.S.C. 101 because the claimed invention is
directed to an abstract idea without significantly more.
Claim 1 recites: (Step 2a, Prong One)
extracting data from a document to build a formulation database.
The limitation of extracting data from a document to build a formulation database, as drafted,
is a process that, under its broadest reasonable interpretation, covers performance of the
limitation in the mind but for the recitation of generic computer components. That is, other than
reciting a generic processors/device, nothing in the claim element precludes the step from
practically being performed in the mind. For example, but for the processors language,
“extracting” in the context of this claim encompasses the user manually determining
a generic “formulation database” using a generic “documents” with data and generic “extracting” steps. Similarly, the limitation(s) of generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. For example, but for the processors language, generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding in the context of this claim encompasses the user manually receiving generic “documents” and “data” and performing generic “sorting” and “extracting” steps. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas (concepts performed in the human mind (including an observation, evaluation, judgment, opinion)).
Further, these concepts also recite “Certain Methods of Organizing Human Activity”; (such as
commercial or legal interactions (including agreements in the form of contracts; legal
obligations; advertising, marketing or sales activities or behaviors; business relations) where
performing generic exporting steps of generic profiles using generic transformations and
templates is a method of human activity in commercial or legal interactions.
Accordingly, the claim recites an abstract idea.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites one additional element – using a processor with a device/method to perform both the generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding; and extracting steps. The processor with a device/method in both steps is recited at a high level of generality (i.e., as a generic processor performing a generic computer function of “extracting”) such that it amounts no more than mere instructions to apply the exception using a generic computer component.
Accordingly, this additional element does not integrate the abstract idea into a practical
application because it does not impose any meaningful limits on practicing the abstract idea.
The claim is directed to an abstract idea. The claim does not include additional elements that
are sufficient to amount to significantly more than the judicial exception.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of a processor with a device/method to perform both the generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding; and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claim(s) is/are not patent eligible.
Referring to claim 2, (Step 2a, Prong One) this further merely performs an additional abstract
mental step of “extracting text segments in the document; predicting a class related to the formulation for the extracted text segments using an object name recognition model;
and storing a paragraph including a sentence including information including at least one of the composition and property of the formulation”.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites the additional elements of “extracting text segments in the document; predicting a class related to the formulation for the extracted text segments using an object name recognition model; and storing a paragraph including a sentence including information including at least one of the composition and property of the formulation” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of using “extracting text segments in the document; predicting a class related to the formulation for the extracted text segments using an object name recognition model; and storing a paragraph including a sentence including information including at least one of the composition and property of the formulation” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
The claim(s) is/are not patent eligible.
Referring to claim 3, (Step 2a, Prong One) this further merely performs an additional abstract
mental step of “checking the locations of tables in the document using a pretrained
table-transformer model; setting a window of a preset size at the checked locations of the tables; extracting contents of a table included in the window; selecting a table including a content including a keyword related to the formulation from the extracted contents using a
Tesseract OCR model; and storing the selected table”.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites the additional elements of “checking the locations of tables in the document using a pretrained table-transformer model; setting a window of a preset size at the checked locations of the tables; extracting contents of a table included in the window; selecting a table including a content including a keyword related to the formulation from the extracted contents using a
Tesseract OCR model; and storing the selected table” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of using “checking the locations of tables in the document using a pretrained table-transformer model; setting a window of a preset size at the checked locations of the tables; extracting contents of a table included in the window; selecting a table including a content including a keyword related to the formulation from the extracted contents using a Tesseract OCR model; and storing the selected table” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
The claim(s) is/are not patent eligible.
Referring to claim 4, (Step 2a, Prong One) this further merely performs an additional abstract
mental step of “converting selected table images into an HTML format using a table structure recognition model; and converting and storing a table to fit the format of the existing database using an HTML parser”.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites the additional elements of “converting selected table images into an HTML format using a table structure recognition model; and converting and storing a table to fit the format of the existing database using an HTML parser” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of using “converting selected table images into an HTML format using a table structure recognition model; and converting and storing a table to fit the format of the existing database using an HTML parser” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
The claim(s) is/are not patent eligible.
Referring to claim 5, (Step 2a, Prong One) this further merely performs an additional abstract
mental step of “generating composite formulation data from original data using a tabular data generator; comparing a conditional distribution between the original data and the composite formulation data using a discriminator to output a similarity; and sampling composite data based on the similarity”.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites the additional elements of “generating composite formulation data from original data using a tabular data generator; comparing a conditional distribution between the original data and the composite formulation data using a discriminator to output a similarity; and sampling composite data based on the similarity” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of using “generating composite formulation data from original data using a tabular data generator; comparing a conditional distribution between the original data and the composite formulation data using a discriminator to output a similarity; and sampling composite data based on the similarity” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
The claim(s) is/are not patent eligible.
Referring to claim 7, (Step 2a, Prong One) this further merely performs an additional abstract
mental step of “generating, in an actor network, third data for a new compound formulation based on first data for a target property and second data for an existing compound formulation;
assigning a reward score based on a similarity between a property value of the new compound formulation of the third data and the target property; and feeding back, in a critic network, an expected value of how close the new formulation changed compared to an existing formulation has improved to the target property based on the reward score to the actor network”.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites the additional elements of “generating, in an actor network, third data for a new compound formulation based on first data for a target property and second data for an existing compound formulation; assigning a reward score based on a similarity between a property value of the new compound formulation of the third data and the target property; and feeding back, in a critic network, an expected value of how close the new formulation changed compared to an existing formulation has improved to the target property based on the reward score to the actor network” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of using “generating, in an actor network, third data for a new compound formulation based on first data for a target property and second data for an existing compound formulation; assigning a reward score based on a similarity between a property value of the new compound formulation of the third data and the target property; and feeding back, in a critic network, an expected value of how close the new formulation changed compared to an existing formulation has improved to the target property based on the reward score to the actor network” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
The claim(s) is/are not patent eligible.
Referring to claim 8, (Step 2a, Prong One) this further merely performs an additional abstract
mental step of “a memory that stores one or more instructions; and a processor that executes the one or more instructions stored in the memory, wherein the processor executes the one or more instructions”.
(Step 2a, Prong Two)
This judicial exception is not integrated into a practical application. In particular, the claim only
recites the additional elements of “a memory that stores one or more instructions; and a processor that executes the one or more instructions stored in the memory, wherein the processor executes the one or more instructions” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
(Step 2b)
The claim does not include additional elements that are sufficient to amount to significantly
more than the judicial exception. As discussed above with respect to integration of the abstract
idea into a practical application, the additional element of using “a memory that stores one or more instructions; and a processor that executes the one or more instructions stored in the memory, wherein the processor executes the one or more instructions” steps to perform both the aforementioned generating; predicting; optimizing; extracting; extracting; checking; sorting; extracting; extracting; extracting; collecting; storing; evaluating; determining and adding and extracting steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
The claim(s) is/are not patent eligible.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-2, 5, 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over
Hong et al. US Pub. No. 2022/0397886 A1, in view of Dorsett et al., US Pub. No. 2002/0128734 A1, in view of Asahara et al., US Pub. No. 2022/0358438 A1, in view of Kakuda et al., US Pub. No. 2024/0378355 A1.
As to claim 1, Hong discloses
an operating method, performed by one or more processors of an electronic device,
(Hong [0001, 0063, 0096])
the operating method comprising:
extracting data from a document to build a formulation database;
(Hong teaches collecting/receiving input chemical material/chemical formulation information dataset including chemical material Information for a database that stores the chemical material data, i.e. “extracting data from a document to build a formulation database” see [0051] The data reception unit 110 may collect or receive a input chemical material dataset including chemical material information, chemical composition information, chemical formulation information, and/or chemical property information
of the chemical formulation. In the input chemical material dataset, information other than the above-mentioned information may be added when the information is required for deriving a chemical formulation, or a part of the above-mentioned information may be excluded. The input
chemical material dataset can be reduced in dimension and complexity of input data by weighting a principal variable or a principal component using feature importance or principal
component analysis (PCA).
see also [0060] The database 170 may store at least one or more of the first chemical material data, the predicting model, the validation model, and the chemical formulation data of the
first to third groups, and further store other information including various parameters generated in an automated chemical formulation process.;
see also [0009] a data reception unit that receives a first chemical material dataset including chemical material information, chemical composition information, chemical formulation information and property information thereof; )
generating composite formulation data based on the formulation database;
(Hong also teaches chemical composition information/ chemical formulation information, i.e. generating composite formulation data see [0009] a data reception unit that receives a first chemical material dataset including chemical material information, chemical composition information, chemical formulation information and property information thereof;)
predicting, using a formulation property prediction model that infers a property change according to the composition of a formulation, the property of a compound from the formulation information of the compound based on the composite formulation data;
(Hong teaches a predicting model/ a formulation prediction unit for target properties , i.e. “predicting, using a formulation property prediction model that infers a property change” [0010] a predicting model generation unit that trains a first machine learning model using the input chemical material dataset to generate a predicting model for predicting a chemical formulation based on target property information of a target material;
[0011] a formulation prediction unit that sets a boundary condition based on the first chemical material dataset, generates a new input dataset including the chemical material information, chemical composition information, and chemical formulation information within the
boundary condition, inputs the new input dataset to the predicting model, and sets predetermined one or more pieces of target property information to perform prediction,
thereby outputting chemical formulation data of a first group;
see also [0052] The predicting model generation unit 120 may train a first machine learning model using the input chemical material dataset to generate a predicting model for predicting
a chemical formulation based on target property information of a target material. Here, a "target material" refers to a chemical material exhibiting a physical property specified by a user. A "target property" refers to one or more features required for a chemical material to be obtained by the user, and may be one or more of physical and/or chemical features. )
and
optimizing a formulation using a formulation optimization model that generates a new compound formulation suitable for a target property in a compound formulation,
(Hong [0052] The predicting model generation unit 120 may train a first machine learning model using the input chemical material dataset to generate a predicting model for predicting
a chemical formulation based on target property information of a target material. Here, a "target material" refers to a chemical material exhibiting a physical property specified by a user. A "target property" refers to one or more features required for a chemical material to be obtained by the user, and may be one or more of physical and/or chemical features.)
Hong does not disclose:
wherein the building of the formulation database comprises:
extracting a paragraph related to a formulation from text included in the document; and
extracting data related to the formulation from a table included in the document, and
wherein the extracting of data related to the formulation comprises:
checking a row having composition information or property
information in a table included in at least one of an HTML document and an XML document;
sorting example numbers assigned to classify experimental examples in a table head, the composition information, and the property information, respectively, to generate a composition data set and a property data set;
extracting a pair of composition information and property information whose example numbers match from the composition data set and the property data set;
extracting information including the name, unit, and amount of a raw material from the pair of the composition information and the property information;
extracting information including the name, value, unit, property experimental method, and property experimental condition of a property from the pair of the composition information and the property information; and
automatically adding information including the name, unit, and amount of the raw material and information including the name, value, unit, property experimental method, and property experimental condition of the property to an existing database;
However, Dorsett discloses:
wherein the building of the formulation database comprises:
extracting a paragraph related to a formulation from text included in the document;
(Dorsett teaches extracting object descriptions, i.e. a paragraph related to formulation see [0047] As described above, client processes 140 receive or generate data derived from an experiment and package that data using known techniques in a format (e.g., an XML data stream) for communication to database server process 130. Upon receiving such a communication from a client process 140, database server process 130 parses the incoming data stream to extract object descriptions from the data stream.;
See also [0020] FIGS. 4A-4B illustrate a portion of an XML document defining a data object representing an experimental data set and a relational database table into which such an
object is mapped, respectively.)
and
extracting data related to the formulation from a table included in the document,
(Dorsett teaches extracting objects/entities, i.e. extracting data related to the formulation from a table see [0038] Database server process 130 parses the data, extracting for each object represented in the XML stream an
object type and optionally one or more associated object properties (step 250), as will be described in more detail below.; see also [0012] Parsing the representation can
include identifying each of a plurality of XML entities in an XML stream, each entity having associated content; mapping each XML entity into a corresponding object property;
and assigning the content associated with an XML entity to a database table based on the corresponding object property.)
and
wherein the extracting of data related to the formulation comprises:
checking a row having composition information or property information in a table included in at least one of an HTML document and an XML document;
(Dorsett teaches mapping rows/columns properties to the database, i.e. “checking a row having composition information or property information” see [0048] In one implementation,
database server process 130 maps classes (e.g., the Experiment class and any subclasses, the Element class, etc.), to database tables, with each row representing an individual instance of the class or classes in the table (e.g., an experiment identified by a unique ExperimentID) and columns corresponding to class properties (e.g., for an experiment, the individual ExperimentID primary key, project name, experiment name, notebook number or the like, although not
all properties need necessarily be saved in database 180))
sorting example numbers assigned to classify experimental examples in a table head, the composition information, and the property information, respectively,
(Dorsett teaches extracting ordering see [0071] In particular, the presence or absence and ordering of entities in the XML document can represent the ordering of (and hierarchies in)
the experimental procedure;
see also [0064] 2. The ability to use the Experiment object itself to determine and represent the time-ordered sequence of all laboratory experiments related to a combinatorial library.)
to generate a composition data set and a property data set;
(Dorsett teaches experiments/composition data and various properties data sets see
[0043] Experiment objects representing ( e.g., referring to) the library. Optionally, experiments that result in reformulation or other alteration of the library ( e.g., experiments that
would result in a significantly different composition of matter and/or phase) can result in the creation of a new library.;
see also [0043] In addition to properties inherited from the Experiment base class, a Synthesis object has additional properties denoting the library's geometry (e.g., the number of rows
and columns in the matrix representing the library of materials), and may also have properties corresponding to the methods used to prepare the library, such as data for reagents, processing conditions and the like.;
see also [0040] When it has received the necessary information, database server process 130
creates an Experiment object representing the preparation of the library-e.g., a Synthesis object, as described below. Optionally, database server process 130 can also create a separate object representing each new library (such as, for example, a Library object having properties specifying, e.g., LibraryID, library geometry, the identity of materials making up the library and the like).)
extracting a pair of composition information and property information whose example numbers match from the composition data set and the property data set;
(Dorsett teaches extracting experiments IDs/experiment objects and matching experiment object data, i.e. “composition information and property information whose example numbers match from the composition data set and the property data set”
See [0051] Assuming the user "jsmith" has appropriate system authorization, after parsing this object in step 250, database server process 130 assigns an Experiment ID and Keyword
IDs and in step 260 stores the information as follows (in EXPERIMENT and KEYWORD tables);
See also [0065] The first benefit means that since the Experiment object contains attributes common to all experiments (such as library ID, date and time, notebook number and page,
staff member, and keywords), it is possible to simply query the database for all experiments performed on a given library (for example) and immediately produce a list.;
see also [0092] Database server process 130 searches the appropriate tables in database 180 for any record that satisfies the specified search terms, and assembles the results into the form of a list object (step 780). The list object can subsequently be used by user interface
program 160 to, for example, present the number of hits returned by the user's query, and it can be independently stored in database 180 for sharing with other users or for later use.)
extracting information including the name, unit, and amount of a raw material from the pair of the composition information and the property information;
(Dorsett [0065] The first benefit means that since the Experiment object contains attributes common to all experiments (such as library ID, date and time, notebook number and page,
staff member, and keywords), it is possible to simply query the database for all experiments performed on a given library (for example) and immediately produce a list;
see also [0032] Laboratory data management system 100 is configured to manage data generated during the course of the experiments performed by laboratory apparatus 150. An
experiment is performed on a library of materials. As used in this specification, a library of materials is a matrix having two or more members, generally containing some variance in chemical or material composition, amount, reaction conditions, and/or processing conditions. A member, in turn, represents a single constituent, location, or position in a library containing one set of chemicals or materials subject to one set of reaction or processing conditions.)
extracting information including the name, value, unit, property experimental method, and property experimental condition of a property from the pair of the composition information and the property information;
(Dorsett [0065] The first benefit means that since the Experiment object contains attributes common to all experiments (such as library ID, date and time, notebook number and page,
staff member, and keywords), it is possible to simply query the database for all experiments performed on a given library (for example) and immediately produce a list;
see also [0032] Laboratory data management system 100 is configured to manage data generated during the course of the experiments performed by laboratory apparatus 150. An
experiment is performed on a library of materials. As used in this specification, a library of materials is a matrix having two or more members, generally containing some variance in chemical or material composition, amount, reaction conditions, and/or processing conditions. A member, in turn, represents a single constituent, location, or position in a library containing one set of chemicals or materials subject to one set of reaction or processing conditions.)
and
automatically adding information including the name, unit, and amount of the raw material and information including the name, value, unit, property experimental method, and property experimental condition of the property to an existing database
(Dorsett teaches creating new libraries in a database with many different experiment properties/condition
See [0043] In one implementation, each library 300 is represented by at least one Experiment object 330, including a Synthesis object (instantiated from a Synthesis subclass of the Experiment base class) reflecting the library's preparation. In addition to properties inherited from the Experiment base class, a Synthesis object has additional properties denoting the library's geometry (e.g., the number of rows and columns in the matrix representing the library of materials), and may also have properties corresponding to the methods used to prepare the library, such as data for reagents, processing conditions and the like. Additional
experiments performed on the same library can result in multiple Experiment objects representing ( e.g., referring to) the library. Optionally, experiments that result in reformulation
or other alteration of the library ( e.g., experiments that would result in a significantly different composition of matter and/or phase) can result in the creation of a new
library.
[0032] Laboratory data management system 100 is configured to manage data generated during the course of the experiments performed by laboratory apparatus 150. An
experiment is performed on a library of materials. As used in this specification, a library of materials is a matrix having two or more members, generally containing some variance in chemical or material composition, amount, reaction conditions, and/or processing conditions. A member, in turn, represents a single constituent, location, or position in a library containing one set of chemicals or materials subject to one set of reaction or processing conditions.).
It would have been obvious to one having ordinary skill in the art at the time the time of the effective filing date to apply experiment library creation using experiment objects as taught by Dorsett to the system of Hong since it was known in the document processing art that it is useful to use an experiment object as a base class for all experiments where this provides several benefits, including: object-oriented design benefits of improving reusability in the sense that this base class identifies the common attributes for all laboratory experiments and the ability to use the Experiment object itself to determine and represent the time-ordered sequence of all laboratory experiments related to a combinatorial library where this means that since the Experiment object contains attributes common to all experiments (such as library ID, date and time, notebook number and page, staff member, and keywords), it is possible to simply query the database for all experiments performed on a given library (for example) and immediately produce a list and where this provides that since the actual name of the specific Experiment-derived class is stored using of the ClassName attribute in the SaveXML method as described above, it is possible to dynamically retrieve the specific experimental detail for any type of experiment by simply inspecting the value of the Experiment.ClassName property retrieved and using that name in a subsequent GetObject request where this permits the construction of universal user software that can view any experimental data records from the past, present and future definitions of experiment objects (Dorsett [0062-0066]).
Hong/Dorsett do not disclose:
by performing selecting only
composition and/or property pairs with matching example numbers from a composition table
and/or a property table, and automatically adding the selected composition and/or property pairs
to the existing database, and
wherein the predicting of the property of the compound comprises:
collecting learning data including original data and the composite formulation data;
evaluating a prediction result for each predicted property, by performing evaluating the
prediction result using a first performance evaluation index for a first predicted property
corresponding to a continuous variable, and evaluating the prediction result using a second
performance evaluation index different from the first performance evaluation index for a first
predicted property corresponding to a categorical variable
however, Asahara discloses:
by performing selecting only composition and/or property pairs with matching example numbers from a composition table and/or a property table,
(Asahara teaches generating cross-task compatible feature values using the experimental/material data table/DB , i.e. “performing selecting only composition and/or property pairs with matching example numbers from a composition table and/or a property table”
See also [0063] Upon receiving the above command to execute interpolation from the material property prediction presenting unit (116), the material property predicting unit (113)
retrieves the data of the experimental data table specified by the task ID (700) from the material DB (112) (S1002). Also, in the screen (1104) in FIG. 11, any other task is selected that is used to generate cross-task compatible feature values. The material property predicting unit (113) retrieves a predictive model (902) related to the selected other task from the material property predictive model DB (114) (S1003).;
see also Fig. 4 item S401: “RECEIVE EXPERIMENTAL DATA ACCEPTING DATA AND RECOGNIZE TASK ID”
see also [0049] FIG. 7 illustrates information in one record of an experimental data table. This data includes experiment ID (701) assigned to each experiment in a serial numbering
scheme or the like so that the experiment can be identified uniquely, material property (702) derived from the material property (601) of the experimental data (600), material
structural formula (703) derived from the material structural formula (602) of the experimental data (600), and experimental condition (704) derived from the experimental condition
(603). Information from which these items of data are derived may be converted in units and formats and transformed into a coherent representation.)
and automatically adding the selected composition and/or property pairs to the existing database,
(Asahara teaches determining cross-task compatible feature values and adding values to the database See [0066] Then, the material property predicting unit (113) generates data for predicting material properties (S1004). This processing corresponds to predicting the material property A by applying the structural formulas in the data of the past task B (903) to the predictive model (902) and adding the material property A to the data of the past task B, thus generating a new data set (904). At this time, the cross-task compatible feature value generating unit (115) executes prediction of the material property A ( cross-task compatible
feature values) by using the predictive model (902) retrieved in the pressing step (S1003).
See also [0057] The process inputs data (item No. 4) for which the material property B should be predicted to the generated predictive model (905) and obtains the material property B.
By adding the material property A as new feature values (cross-task compatible feature values), it can be expected to improve the prediction accuracy in comparison with when the past task B data is used as it is. This is considered as effective particularly when there is a correlation between the material properties A and B.
See also <2. Material Data Inputting Process> [0044] FIG. 4 illustrates an example of a processing procedure of the material DB update process (S311). In the material DB update process (S311), the experimental data accepting unit (111) first receives experimental data (600) from the user (102) and recognizes or adds a task ID (S401). Then, the process updates or adds the corresponding data per task to the material DB (112) (S402).;
See also Fig. 4 item S402: “ADD CORRESPONDING DATA PER TASK TO MATERIAL DB” MATERIAL DB (112) )
and
wherein the predicting of the property of the compound comprises:
collecting learning data including original data and the composite formulation data;
(Asahara teaches collecting experimental data for the predictive model, i.e. “collecting learning data”
See [0048] The first step (S401) of the material DB update process (S311) of FIG. 4 interprets and formats the experimental data (600) and stores it as an experimental data table
in the material DB (112).
See also [0013] The material property predicting unit generates the first predictive model by using the material compositions, experimental condition, feature values, and the known material property in the first task data. Also, the material property predicting unit inputs the material composition, the experimental condition, and the feature value in a record in which the material property is unknown in the first task data to the first predictive model and predicts the unknown material property.;
[0075] Although structural formulas are used when creating feature values about any other task in the example discussed hereinbefore, data of composition and others may be used as long as the data is common across tasks data. Additionally, a method in which prediction can be made using structural formulas as such is also publicly known and the scheme is the same in that case as well.).
evaluating a prediction result for each predicted property, by performing evaluating the
prediction result using a first performance evaluation index for a first predicted property
corresponding to a continuous variable, and evaluating the prediction result using a second
performance evaluation index different from the first performance evaluation index for a first
predicted property corresponding to a categorical variable
(Asahara teaches material property predictive model DB using explanatory variables and the material property, i.e. using performance evaluation indexes see [0069] From the data for predicting material properties except for records in which material property (702) is unmeasured, i.e., blank, the material property predicting unit (113) assigns items excepting task ID (700), experiment ID (701), and material property (702) to the explanatory variables and the material property (702) to the objective variable, executes a regression analysis which is publicly known, obtains a prediction function, and learns a predictive model (905) (S1005). The created predictive model (905) is stored into the material property predictive model DB (114) together with the task ID of the data from which the predictive model (905) was generated.;
See also Fig. 9 showing multiple task indexes)
It would have been obvious to one having ordinary skill in the art at the time the time of the effective filing date to apply material property predicting and cross-task compatible feature values as taught by Asahara to the system of Hong / Dorsett since it was known in the machine learning/material prediction processing art that upon having obtained the new data set the process generates a predictive model to predict a material property B, taking known data of the material property B in the data set as teacher data when at this time, the explanatory variables are the structural formulas, experimental condition (humidity), and the material property A and the objective variable is the material property B where the predictive model can be generated through supervised machine leaning which is known and the process inputs data for which the material property B should be predicted to the generated predictive model and obtains the material property B and by adding the material property A as new feature values
(cross-task compatible feature values), it can be expected to improve the prediction accuracy in comparison with when the past task B data is used as it is where this is considered as effective particularly when there is a correlation between the material properties A and B. (Asahara [0056-0057]).
Hong/Dorsett/Asahara do not disclose:
storing a weight value of a model for each property that has been trained through
knowledge transfer, and carrying out performance evaluation on untrained data;
determining which of a plurality of preset classifications corresponds to table data in the
document, and applying the determined preset classification;
However, Kakuda discloses:
storing a weight value of a model for each property that has been trained through
knowledge transfer, and carrying out performance evaluation on untrained data;
(Kakuda abstract: A property prediction device includes a processor; and a memory storing program instructions that cause the processor to: create a prediction model by using a training dataset of a composite material including raw materials in first and second categories to perform machine learning of a correspondence relationship between a property of the composite
material, which is an objective variable, versus a mixing amount of the raw material in the first category and a weighted feature of the raw material in the second category, which are explanatory variables; see also [0013] a prediction part configured to input, as explanatory
variables, a mixing amount of a raw material in the first raw material category and a weighted feature of a raw material in the second raw material category, that are created based on prediction data of a composite material whose property is to be predicted, into the
prediction model so as to predict the property of the composite material corresponding to the prediction data.)
and
determining which of a plurality of preset classifications corresponds to table data in the
document, and applying the determined preset classification;
(Kakuda [0104] FIG. 13 is a diagram illustrating an example of a configuration of explanatory variables. The explanatory variables of FIG. 13 are created by combining weighted
features in the "curing agent" category selected as the raw material category to be optimized on the screen 1000 of FIG. 12, with the mixing amounts in raw material categories not
selected as the raw material category to be optimized. As illustrated in FIG. 13, the information processing system 1 according to the present embodiment can create the explanatory variables by combining elements represented as the mixing amounts of raw materials in the raw material categories not selected as the raw material category to be optimized, with elements represented as weighted features of raw materials of the raw material category selected as the raw material category to be optimized, and can use the created explanatory variables for machine learning.
See also [0008] In a property prediction device for predicting a property of a composite material composed of raw materials in a plurality of raw material categories, a need exists for a
technology by which, when a new raw material in a specific raw material category such as an additive is searched, the influence of the specific raw material category on the property of the composite material can be accurately predicted)
It would have been obvious to one having ordinary skill in the art at the time the time of the effective filing date to apply categories/material property predicting as taught by Kakuda to the system of Hong / Dorsett / Asahara since it was known in the machine learning/material prediction processing art that information processing system can create explanatory variables by calculating weighted features of raw materials in a specific raw material category, and combining the weighted features with the mixing amounts of raw materials in another raw
material category where the influence of the other material category can be reduced, and the influence of the specific raw material category on the property of the composite material can be accurately predicted where the information processing system according to the present embodiment can accurately predict a change in a property of a composite material when the
molecular structure of an additive (addition) having a smaller mixing amount than that of a main raw material is changed and in addition, in order to satisfy a desired property, additives to be used can be screened where the information processing system where this is particularly effective when some of substructures of an additive and a main raw material are the same. (Kakuda [0113-0114]).
As to claim 2, Dorsett as modified discloses the operating method of claim 1, wherein the extracting of a paragraph related to the formulation comprises:
extracting text segments in the document;
(Dorsett teaches extracting text strings, i.e. extracting text segments
See [0047] Upon receiving such a communication from a client process 140, database server process 130 parses the incoming data stream to extract object descriptions from the
data stream.
see also [0046] Each object has a set of properties that can include, e.g., object metadata, attributes and joins…. Object attributes include the set of properties that can be assigned values for any given instance of the object Each attribute may be described, e.g., by name, description, the type of data the attribute stores (for example integer data, floating point, text strings, image, or x-y data),)
predicting a class related to the formulation for the extracted text segments using an object name recognition model;
(Dorsett teaches generating new derived classes/class names, i.e. predicting a class see [0088] Thus in the case of a "new experiment type" discussed above, upon encountering an unrecognized classname database server process 130 can be configured to
create an instance of the Experiment base class describing the general characteristics of the experiment as discussed above, and to store the as-yet-unrecognized experimental detail embodied in the particular derived properties in generic storage, for later explicit treatment by a system administrator or the like. Alternatively, database server process 130 can be configured to generate new derived Experiment classes dynamically, using information contained
within the XML representation as a framework for identifying and populating object properties, and/or mapping such properties to new or existing tables in database 180.)
and
storing a paragraph including a sentence including information including at least one of the composition and property of the formulation
(Dorsett [0093] As discussed above, in one implementation database server process 130 formats the recordset as a list of elements that satisfy the terms of the query. Lists can be
stored in database 180 and can be exported to files for storage. To store a list in database 180, the user selects File>Save List from a menu and selects the desired list from a set of available lists 850 (including, e.g., one or more lists generated by the user during a particular session). Optionally, user interface program 160 can prompt the user for list metadata, such as a flag indicating that the list should be public (i.e., made available to other users of system 100) or
private, or a project name and/or description to be associated with the stored list. In one implementation, stored lists are static (i.e., database server process 130 does not update the
list to include query-satisfying records added to database 180 after the underlying query was saved), while database server process 130 updates the data content of records included in a stored list when the list is retrieved from database 180 or opened by user interface program 160.).
As to claim 5, Hong as modified discloses the operating method of claim 1, wherein the generating of the composite formulation data comprises:
generating composite formulation data from original data using a tabular data generator;
(Hong [0059] The validation unit 160 may compare and validate the first group of chemical formulation data with the second group of chemical formulation data, and select and derive
matching data as chemical formulation data of a third group corresponding to the predetermined one or more pieces of target property information.)
comparing a conditional distribution between the original data and the composite formulation data using a discriminator to output a similarity;
(Hong [0059] The validation unit 160 may compare and validate the first group of chemical formulation data with the second group of chemical formulation data, and select and derive
matching data as chemical formulation data of a third group corresponding to the predetermined one or more pieces of target property information. ; see also [0072] In step S270, the first group of chemical formulation data is compared and validated with the second group of chemical formulation data , and matching data is selected and derived as chemical formulation data of a third group suitable for the predetermined one or more pieces of target property information. The chemical formulation solutions in the third group of chemical formulation data may be ranked in suitability according to a predetermined condition)
and
sampling composite data based on the similarity
(Hong [0086] With given input vectors, the goal of the SVM is to find wand b so that sign(wT(D(x)+b) is accurate for most samples.).
As to claim 8, Hong as modified discloses an electronic device that performs the method of claim 1, the electronic device comprising:
a memory that stores one or more instructions; and a processor that executes the one or more instructions stored in the memory, wherein the processor executes the one or more instructions
(Hong [0001, 0050, 0037, 0063, 0096]).
Claim(s) 3-4 is/are rejected under 35 U.S.C. 103 as being unpatentable over
Hong et al. US Pub. No. 2022/0397886 A1, in view of Dorsett et al., US Pub. No. 2002/0128734 A1, in view of Asahara et al., US Pub. No. 2022/0358438 A1, in view of Kakuda et al., US Pub. No. 2024/0378355 A1, in view of Gu et al. US Pub. No. 2022/0382975 A1, in view of Rodriguez et al., US Pub. No. 2020/0401798 A1.
As to claim 3, Hong/Dorsett/Asahara/Kakuda do not disclose:
wherein the extracting of data related to the formulation comprises:
checking the locations of tables in the document using a pre- trained table-transformer model;
setting a window of a preset size at the checked locations of the tables;
extracting contents of a table included in the window;
selecting a table including a content including a keyword related to the formulation from the extracted contents using a Tesseract OCR model; and
storing the selected table;
However, Gu discloses:
the operating method of claim 1, wherein the extracting of data related to the formulation comprises:
checking the locations of tables in the document using a pre- trained table-transformer model;
(Gu teaches a pre-trained document detector extracts semantically-meaningful components that including tables see [0071] In some examples, pre-training the machine learning model 710
includes detecting the following categories: blocks of text, titles, lists, tables, and figures within documents 702 to obtain detected document proposals.;
see also [0047] In one example, training the machine learning model 110 includes obtaining pre-training data that includes multiple documents (e.g., a document dataset). In some cases, the documents include a public document dataset. These documents are fed to document detector 112, which detects objects within each of the documents of the document dataset. For instance, the document detector 112 identifies semantically-meaningful components such as blocks of text, a title, lists, tables, figures, etc. The document detector 112 localizes these semantically-meaningful components using bounding boxes to provide annotations. The document detector 112 extracts semantically-meaningful components that include visual features.)
setting a window of a preset size at the checked locations of the tables;
(Gu teaches various bounding boxes, i.e. “windows” see [0025] Some existing computer-based methods utilize graphical properties of documents, employing convolutional neural networks to recognize a bounding box and semantic segmentation in a document.; see also [0048] Additionally, the document detector 112 sends a proposed document representation to the OCR engine 114. The proposed representation includes visual features, which are each identified in a bounding box, cropped bounding box, bounding capsule, or minimum bounding rectangle (MBR), etc.).)
extracting contents of a table included in the window;
(Gu [0064] At block 406, the process 400 involves applying an OCR engine to each of the
documents within the set of documents. At block 408, the process 400 involves extracting textual features and visual features from each of the documents within the set of documents.;
see also [0066] At block 502, the process 500 involves accessing training data that includes a set of documents for pretraining a machine learning model 110. At block 504, the process 500 involves extracting textual features and visual features from each of the documents within the set of documents.)
selecting a table including a content including a keyword related to the formulation from the extracted contents using a Tesseract OCR model;
(Gu [0100] The machine learning model 110 uses the document detector 112 to generate RoI heads by cropping each of the proposals from the original document images and expanding each of the bounding boxes by a factor of 1.1. After obtaining detection results from the document detector 112, an OCR engine (e.g., OCR engine 114, a public OCR engine, or a Tesseract OCR, etc.) extracts plain text from each proposal obtained from the document detector 112.)
It would have been obvious to one having ordinary skill in the art at the time the time of the effective filing date to apply pre-trained document detectors as taught by Gu to the system of Hong/Dorsett/Asahara/Kakuda since it was known in the document processing art that document processing systems provide a machine learning model which performs document pre-processing on the pre-training data where the training document detector involves using a model with rotations on images are applied as a form of data augmentation to improve an overall quality of detection of
potential vertical text features in documents where the document detector generates 2D proposals for visual features in the documents, for example, by applying bounding boxes to
each of the proposals where the machine learning model uses the document detector to generate RoI heads by cropping each of the proposals from the original document images and expanding each of the bounding boxes and after obtaining detection results from the
document detector, an OCR engine (e.g., OCR engine, a public OCR engine, or a Tesseract OCR, etc.) extracts plain text from each proposal obtained from the document detector
(Gu [0100]).
Hong/Dorsett/Gu do not disclose:
and
storing the selected table;
However, Rodriguez discloses:
and
storing the selected table
(Rodriguez teaches storing generating queryable data structures for tables, i.e. storing the selected table [0064] The preferred embodiment described above can provide an exceptionally efficient, generic process for generating queryable data structures for tables. The process
accommodates multiple hierarchy levels in rows and columns, and can detect structure without relying solely on presence of lines or assuming any particular table format.;
see also [0047] The data structure 25 generated by the above method provides a queryable representation of the original table, and can be used for automated data extraction as
indicated in FIG. 5. Step 50 here represents receipt by computer 1 of a request for information which was contained in the original table. Such a request may comprise a standard database query in any convenient format, and may identify data required from the table as a function of the headers, e.g. requesting all or specified subset of data corresponding to one or more specified headers.).
It would have been obvious to one having ordinary skill in the art at the time the time of the effective filing date to apply storing queryable representations of tables as taught by Rodriguez to the system of Hong/Dorsett/Asahara/Kakuda/Gu since it was known in the art that document processing systems provide for a self-supervised document representation learning that provides a task-agnostic pre-training framework for document image analysis by establishing representation learning at a semantic-component level, instead of for a single word or a single character in documents where by feature embedding on document components, the task-agnostic pre-training framework avoids excessive contextualized learning between every word in a document, while still mining relationships between each semantic component. Further, the self-supervised document representation learning described herein introduces cross-modality learning during a pre-training phase for contextualized comprehension on document components across language and vision and as a result, the self supervised document representation learning described herein more effectively leverage multimodal information
from document images, for example, without a need for more costly documents that already include annotations. (Rodriguez [0029]).
As to claim 4, Dorsett as modified discloses the operating method of claim 3, wherein the extracting of data related to the formulation further comprises:
converting selected table images into an HTML format using a table structure recognition model;
(Dorsett [0086] In one implementation, client processes 140 interact with the underlying data through a proxy object server, such as a COM dll, configured to receive an XML representation of data from database server process 130 and construct an set of interfaces (consistent, e.g., with Microsoft's COM standard or the COREA standard) that present a set of methods
and properties representing a particular object or objects. Alternatively, client processes 140 can interact with the data directly through the XML representation-for example, by processing the XML document using a set of XSLT transformation rules to generate an HTML document which is then presented to the user in a Web browser as an Experiment write-up document).
and
converting and storing a table to fit the format of the existing database using an HTML parser
(Dorsett [0086] In one implementation, client processes 140 interact with the underlying data through a proxy object server, such as a COM dll, configured to receive an XML representation of data from database server process 130 and construct an set of interfaces (consistent, e.g., with Microsoft's COM standard or the COREA standard) that present a set of methods
and properties representing a particular object or objects. Alternatively, client processes 140 can interact with the data directly through the XML representation-for example, by processing the XML document using a set of XSLT transformation rules to generate an HTML document which is then presented to the user in a Web browser as an Experiment write-up document).
Claim(s) 7 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hong et al. US Pub. No. 2022/0397886 A1, in view of Dorsett et al., US Pub. No. 2002/0128734 A1, in view of Asahara et al., US Pub. No. 2022/0358438 A1, in view of Kakuda US Pub. No. 2024/0378355 A1, in view of Horwood et al., WO 2021/044365 A1.
As to claim 7, Hong/DorsettAsahara/Kakuda do not disclose:
the operating method of claim 1, wherein the optimizing of the formulation comprises:
generating, in an actor network, third data for a new compound formulation based on first data for a target property and second data for an existing compound formulation;
assigning a reward score based on a similarity between a property value of the new compound formulation of the third data and the target property; and
feeding back, in a critic network, an expected value of how close the new formulation changed compared to an existing formulation has improved to the target property based on the reward score to the actor network
However, Horwood discloses
the operating method of claim 1, wherein the optimizing of the formulation comprises:
generating, in an actor network, third data for a new compound formulation based on first data for a target property and second data for an existing compound formulation;
(Horwood [0156] In one or more embodiments of the molecule generation procedure 300,
the options-critic architecture 450 is used to select transformations 420 comprising
15 chemical reactions, and associated reactants 430.; see also [0151] In one or more embodiments, the molecule generation procedure 300 is configured to use deep reinforcement learning algorithms, such as, but not limited to: actor-critic architecture, option-critic architecture, and the like. [0152] Figure 4A depicts a schematic diagram of an actor-critic architecture 400
of the molecule generation procedure 300 in accordance with non-limiting embodiments
15 of the present technology.)
assigning a reward score based on a similarity between a property value of the new compound formulation of the third data and the target property; and
(Horwood [0132] Reward Distribution In the framework of the present technology, the separation between the agent 320 and the environment 340 enables to maintain property-focused rewards that guide optimization while ensuring chemical constraints are met via environment design.
In one or more embodiments, a deterministic reward function may be used based on the
property to be optimized. In order to avoid artificially biasing the agent 320 towards
greedy policies, intermediate rewards are removed and provide evaluative feedback only
at termination of an episode. It is contemplated that using an intermediate reward
discounted by a decreasing function of the step t may improve learning efficiency)
feeding back, in a critic network, an expected value of how close the new formulation changed compared to an existing formulation has improved to the target property based on the reward score to the actor network
(Horwood [0132] Reward Distribution In the framework of the present technology, the separation between the agent 320 and the environment 340 enables to maintain property-focused rewards that guide optimization while ensuring chemical constraints are met via environment design.
In one or more embodiments, a deterministic reward function may be used based on the
property to be optimized. In order to avoid artificially biasing the agent 320 towards
greedy policies, intermediate rewards are removed and provide evaluative feedback only
at termination of an episode. It is contemplated that using an intermediate reward
discounted by a decreasing function of the step t may improve learning efficiency;
see also [0019] In one or more embodiments of the method, the method further comprises,
20 prior to said calculating the reward value using the reward function: receiving, from the
database, based on the product state, a set of properties, said calculating the reward value
using the reward function is based on the set of properties.).
It would have been obvious to one having ordinary skill in the art at the time the time of the effective filing date to apply actor-critic architecture and reward functions as taught by Horwood to the system of Hong/Dorsett/Asahara/Kakuda since it was known in the art that machine learning systems provide for selecting, from the database, the action includes the transformation to apply on the current state to obtain the product state is based on the indication of the molecule where the transformation comprises one of: an addition of a molecular fragment, a deletion of a molecular fragment, and a substitution of a molecular fragment where the method further comprises, prior to said calculating the reward value using the reward function: receiving, from the database, based on the product state, a set of properties, said calculating the reward value using the reward function is based on the set of properties. (Horwood [0017-0019]).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Kamesawa et al., JP2021076890A, teaches to provide a compound property predictor capable of predicting the property of a target compound in an optimization program for a lead
compound. SOLUTION: The compound database in which the measured properties of
each of the compounds are associated is accessible, and two compounds selected from
the compound database are used as the selection compound, and the common structure
and the difference structure of the selection compound and the difference structure of
the selection compound are used. It was selected from a property learning means that
was machine-learned to predict the properties of the compound to be predicted, and the
compound to be predicted and the compound database, using it as supervised training
data including at least the combination of the properties. The compound property
prediction device 100 includes a property prediction means for obtaining a prediction
result of the property of the compound to be predicted as an output of the property
learning means by inputting the common structure and the difference structure of the
compound into the property learning means;
Lee al., US Pub. No. 2025/0005451 A1, teaches a composition search method for a material includes constructing a prediction model by learning training data in which information related to a composition of a material is set as an explanatory variable and a value of a physical property of the material is set as an objective variable; calculating a predicted value of the physical property by inputting, into the prediction model, prediction data for newly searching for a composition; calculating an influence degree of each explanatory variable on prediction by using
the training data and the prediction model; calculating a weighted distance of the prediction data with respect to the training data by using the influence degree; and displaying a relationship between the predicted value and the weighted distance, and outputting corresponding prediction data as a search candidate;
Vargas et al., US Pub. No. 2024/0203537 A1, teaches methods include training a machine learning module to predict one or more target product properties for a prospective chemical formulation, including (a) constructing or updating a training data set from one or more variable parameters; (b) performing feature selection on the training data set; ( c) building one or more machine learning models using one or more model architectures; ( d) validating the one or more machine learning models; (e) selecting at least one of the one or more machine learning models and generating prediction intervals; (g) interpreting the one or more machine learning models; and (h) determining if the one or more target product properties calculated are acceptable and deploying one or more trained machine learning models, or optimizing the one or more machine learning models by repeating steps (b) to (g). Methods also include application of trained machine learning modules to predict formulation properties from prospective data.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
CONTACT INFORMATION
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EVAN S ASPINWALL whose telephone number is (571)270-7723. The examiner can normally be reached Monday-Friday 8am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Neveen Abel-Jalil can be reached at 571-270-0474. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Evan Aspinwall/Primary Examiner, Art Unit 2152
Prosecution Insights