DETAILED ACTION
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 .
Priority
Acknowledgement is made of applicant’s claim for priority based on PCT Application No. PCT/CA2023/000030 (filed 26 September 2023), which claims benefit of priority to provisional Application No. 63/465,118 (filed on 9 May 2023) and provisional Application No. 63/409,877 (filed on 26 September 2022).
Information Disclosure Statement
Applicant is reminded of the continuing obligation under 37 CFR 1.56 to timely apprise the Office of any information which is material to patentability of the claims under consideration in this application.
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 5-6 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.
The claims recite converting a tree into a graph database (with claim 6 reciting “converted and incorporated”). However, given that a tree is a data structure, and a database (including graph databases) are systems, it is unclear what is meant by a tree being “converted” into a database, i.e., how a data structure is “converted” into a system (as opposed to, e.g., “stored”). For purposes of examination, the interpretation that “converted” (and “converted and incorporated”) means “stored”, has been taken.
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 1-17 are rejected under 35 U.S.C. 101 because the claims are directed to a judicial exception (i.e., an abstract idea) without significantly more.
Claims 1 and 4 recite preprocessing data to generate and extract information and to generate preprocessed data. This encompasses an evaluation, observation, and/or judgment, which falls under the “Mental Processes” grouping of abstract ideas, as well as managing personal behavior or relationships or interactions between people, which falls under the “Certain Methods of Organizing Human Activity” grouping of abstract ideas.
Claims 2 and 14 recite (repeatedly) decomposing a collection of data into smaller and smaller sub-groups of data, and tokenizing each sub-group of data that the collection of data was decomposed into. Similarly, Claim 3 recites decomposing the collection of data into the smallest units of data for a data type, and tokenizing the smallest units of data. Claim 4 recites repeating the decomposition step until all of the data has been tokenized. These encompass an evaluation, observation, and/or judgment (i.e., determining how/where to decompose the data into sub-groups and even smaller sub-groups, etc.), which falls under the “Mental Processes” grouping of abstract ideas.
Claim 10 recites performing a language-based task. This encompasses managing personal behavior or relationships or interactions between people, which falls under the “Certain Methods of Organizing Human Activity” grouping of abstract ideas.
Claim 11 recites labelling one or more results of (some sort of analysis) to the preprocessed data. This encompasses an evaluation, observation, and/or judgment, which falls under the “Mental Processes” grouping of abstract ideas, as well as managing personal behavior or relationships or interactions between people, which falls under the “Certain Methods of Organizing Human Activity” grouping of abstract ideas.
Because the claims cover performance of the limitation in the mind but for the recitation of generic computer components, the claims fall within the “Mental Processes” grouping of abstract ideas. Similarly, because the claims cover managing personal behavior or relationships / interactions between people but for the recitation of generic computer components, the claims fall within the “Certain Methods of Organizing Human Activities” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
The judicial exception is not integrated into a practical application of the idea. The independent claims recite various computing components, including “graph database” “AI-based model”, and “execution block”. Independent Claim 14 names the various components/modules for performing the disclosed functions. Claims 8-9 and 15-16 have more specificity with respect to the execution block having an execution unit for applying an AI based model to the data. However, these are recited at a high level of generality and recited so generically that they represent no more than mere instructions to apply the judicial exception on a computer (see MPEP 2106.05(f)). These limitations can also be viewed as nothing more than an attempt to generally link the use of the judicial exception to the technological environment of a computer (see MPEP 2106.05(h)).
The claims recite insignificant extra-solution activities, including receiving/retrieving/sending data. See Claims 1 and 7-9. Claim 13 further recites the insignificant extra-solution activity of saving various data in computer readable and computer accessible media. Claims 2-6 recite storing data (i.e., tokens in the tree, and “converting” and “incorporating” the tree into the (graph) database). Claims 3-4 recite “grafting” a resulting tree to a larger tree, which is a form of storing different records in relation to one another (i.e., a form of storing data / recordkeeping).
The claims also variously recite insignificant field-of-use limitations, describing the context rather than a particular manner of achieving the result. These include that the decomposition is performed “recursively” (Claims 2-3), that tokenized data is stored in a tree / tree nodes (Claims 2-4 and 14), that trees are converted (i.e., stored) in a graph database (as opposed to, e.g., a database or other generic storage) (Claims 5-6), that the graph database is “preexisting” (Claim 6), that data is retrieved from a graph database (as opposed to, e.g., a database or other generic storage) and the type of information being retrieved are tokenized (Claim 7), that the execution block comprises an execution unit for carrying out the AI model’s claimed step of being applied to a set of tokenized data, and that the multiple sets of tokenized data are sent to the execution unit in a sequential manner (Claim 8), that the execution block comprises a plurality of execution units operating in parallel, and that the execution block comprises an execution unit for carrying out the AI model’s claimed step of being applied to a set of tokenized data, with the execution occurring simultaneously in parallel (Claims 9 and 16), that it is a language-based task that is being performed (Claim 10), that the language-based task is any one of sentiment analysis, etc. (Claim 12), that the data being saved comprises all processes, data transformations, and data adjustments, and that they are stored in computer readable and computer accessible media (Claims 13 and 17), and that the execution block comprises an execution unit implementing the AI related model on sequentially processed groups of tokenized input data (Claim 15).
As such, the additional elements do not integrate the abstract idea into a practical application of that idea.
With respect to the well-understood, routine, and conventional elements, as stated previously above, the claims do not include any additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract idea into a practical application, the additional elements reciting the use of various computing components amount to no more than mere instructions to apply the judicial exception using generic components. Mere instructions to apply an exception using generic computer components cannot provide an inventive concept (or even components slightly narrower than generic computer components).
Additionally, with regards to the claims’ recitation of receiving and transmitting/sending data, storing (and retrieving) data, and displaying data are well-understood, routine, and conventional activities within the computing realm. See MPEP 2106.05(d)(II) (“Receiving or transmitting data over a network, e.g., using the Internet to gather data” with respect to the receiving/retrieving/sending data steps; “Storing and retrieving data in memory” and “Electronic recordkeeping” with respect to the “storing”, “grafting”, “converting”, and “saving” steps).
Even as an ordered combination, the claims as a whole do not contain any additional elements that amount to significantly more. Claim 1 and its dependent claims do not contain any detail that would further limit the claims to a particular manner by which data is preprocessed. As a result, the claims do not contain any concrete embodiment to that idea, but instead are directed to the resulting goal or effect, rather than a particular manner of achieving such steps.
The presence of the claimed computing elements does not alter this analysis. Simply stating that a computer performs the claimed mental steps, does nothing more than attempt to limit the claims to a particular technological environment. Thus, when removing the computing elements, the claims do no more than describe a desired function or outcome, without providing any limiting detail that confines the claims to a particular solution to an identified problem by a computer aside from invoking the computer as a tool to be used in executing the claimed steps, i.e., applying the abstract idea with a computer. The purely functional nature of the claim confirms that it is directed to an abstract idea, not to a concrete embodiment of that idea (see Affinity Labs of Texas LLC v. Amazon.com Inc., 838 F.3d 1253 (Fed. Cir. 2016) at p. 7-8, citing Elec. Power Grp., LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir. 2016), slip op. 12 (“[T]he essentially result-focused, functional character of claim language has been a frequent feature of claims held ineligible under § 101”)).
As a whole, the claims do not go beyond stating the relevant functions in general terms, without limiting them to a technical means for performing the functions that are arguably an advance over conventional computing technologies. Neither stating an abstract idea while adding the words “apply it” with a computer, nor limiting the use of an abstract idea to a particular technological environment is enough for patent eligibility. Stating the abstract idea while adding the words “apply it with a computer” simply combines those two steps, with the same deficient result.
Therefore, for at least the aforementioned reasons, these claims are rejected under 35 U.S.C. 101 for being directed to a judicial exception (i.e., an abstract idea) without significantly more.
Claim Rejections - 35 USC § 103
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.
Claims 1 and 10-13 are rejected under 35 U.S.C. 103 as being unpatentable over Dawson et al. (“Dawson”) (US 2010/0228693 A1), in view of Sengupta et al. (“Sengupta”) (US 2023/0061731 A1).
Regarding claim 1: Dawson teaches A method for processing a corpus of data …, the method comprising:
a) receiving said corpus of data; b) applying a pre-processing method to said corpus of data to generate and extract information regarding said data in said corpus of data and to generate preprocessed data, said preprocessed data being stored in a … database (Dawson, [0030], where information (in the form of semantic structures) is extracted from resources, which may comprise unstructured and structured data. See also Dawson, [0113], where resources are acquired via a discovery mechanism that utilizes semantic-based searching and web-crawling in the web. With resources in hand, the system then preprocesses text-based documents, specifically identifying text zones, splitting sentences and tokenizing, then performing a step of lexical and terminological analysis, and executing a step of syntactic analysis leading to incipient semantic structures, i.e., concept-relation-concept triples. These are ultimately used to produce one or more semantic structures representing a simple clause associated with each tree; and stores in the memory the one or more semantic structures in association with one or more connections between common ones of the tree elements, thereby generating a set of one or more semantic networks representing the resource, which includes storing one or more resource description framework (RDF) statements comprising RDF elements in association with one or more connections between common ones of the RDF elements. See also Dawson, [0256], where RDF graphs 518 that includes RDF elements, are stored in memory of the system, e.g., a database of the system) … .
Although Dawson does not appear to explicitly state that the database is a “graph database” as claimed, one of ordinary skill in the art would have found it obvious to have modified Dawson to have utilized a graph database with the motivation of storing data that is highly connected to provide flexibility in adding data, running faster relationship-based searches, and indexing relationships.1
Dawson does not appear to explicitly teach [processing a corpus] using at least one AI-based model; [and] c) applying said at least one AI-based model to said preprocessed data using an execution block.
Sengupta teaches [processing a corpus] using at least one AI-based model; [and] c) applying said at least one AI-based model to said preprocessed data using an execution block (Sengupta, [0091-0096], where the significance recognition machine learning model 800 processes word-level representation data objects 802 and character-level representation data objects 804 via one or more attention stages, where multiple attention heads are specifically used to capture multiple different relationships between word-level tokens 602 (i.e., “preprocessed data”), etc., as well as generating a significance token label indicating a name significance subtype 604A for each word-level token. See also Sengupta, [0088] and [0096], where the significance recognition machine learning model processes each word-level token 602 (i.e., “preprocessed data”) to generate an inferred token label, as well as a significance token label indicating a name significance subtype 604A for each word-level token.
See Sengupta, [0006], where the disclosed significant recognition machine learning model may be implemented as a program product comprising program code portions that include execution portions configured to cause the system to implement the disclosed steps; see also, e.g., Sengupta, [0062], where the AI computing entity may retrieve and/or execute one or more of the predefined program algorithms upon the occurrence of a predefined trigger event).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson and Sengupta (hereinafter “Dawson as modified”) with the motivation of enabling the system to learn, interpret, and understand human language in order to apply to a wide range of human-related tasks, e.g., healthcare services as described by Sengupta, or personal knowledge learning, as in Dawson, which enables greater relevancy and useability.
Regarding claim 10: Dawson as modified teaches The method according to claim 1, wherein said Al based model performs a language-based task (Sengupta, [0091-0096], where the significance recognition machine learning model 800 processes word-level representation data objects 802 and character-level representation data objects 804 via one or more attention stages, where multiple attention heads are specifically used to capture multiple different relationships between word-level tokens 602, etc., as well as generating a significance token label indicating a name significance subtype 604A for each word-level token).
Regarding claim 11: Dawson as modified teaches The method according to claim 1, wherein step c) includes applying a step of labelling one or more results from an application of said Al based model to said preprocessed data (Sengupta, [0088] and [0096], where the significance recognition machine learning model processes each word-level token 602 (i.e., “preprocessed data”) to generate an inferred token label, as well as a significance token label indicating a name significance subtype 604A for each word-level token).
Regarding claim 12: Dawson as modified teaches The method according to claim 10, wherein said language-based task is any one of: - sentiment analysis; - relation extraction; - named entity recognition; - conditional generation; - summarization; - predicting a next speaker; - predicting sentiment; - predicting next statements; - symbolic composition (Dawson, [0113], where the system performs extraction and tokenization, executing a step of syntactic analysis leading to incipient semantic structures, i.e., concept-relation-concept triples, implying that relation extraction was performed. See Sengupta, [0041], where the significance recognition machine learning model recognizes, identifies, labels, extractions, and/or the like, significant word-level tokens in textual data objects) .
Regarding claim 13: Dawson as modified teaches The method according to claim 1, further comprising a step of saving all processes and data transformations and data adjustments in computer readable and computer accessible media (Dawson, [0046], where the disclosed system may be carried out as a computer program product that may be provided in any computer-readable form. The semantic network, which is the result of data transformations from extracting data from documents (see, e.g., Dawson, [0088-0089] and [0140-0156]), is stored in memory may be modified (see, e.g., Dawson, [0088-0089] and [0285])).
Claims 2, 14, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Dawson et al. (“Dawson”) (US 2010/0228693 A1), in view of Sengupta et al. (“Sengupta”) (US 2023/0061731 A1), in further view of Fusco et al. (“Fusco”) (US 2023/0055769 A1).
Regarding claim 2: Dawson as modified teaches The method according to claim 1, wherein said pre-processing method includes a decomposition step, said decomposition step comprising recursively decomposing a collection of data from said corpus of data into smaller and smaller sub-groups of data, wherein … a resulting token is stored in a node in a tree along with multiple parameters relating to said resulting token (Dawson, [0140-0156], where the system preprocesses documents to create a document container per document that describes structure and identifies parseable text zones; splits the text into paragraphs, paragraphs into sentences, and employs tokenizers to obtain lexical units (as seen, each of these groups becomes recursively decomposed into smaller and smaller units until tokens, or “lexical units” (i.e., “smallest units of data”) are obtained via tokenizers. These are ultimately used to produce one or more semantic structures representing a simple clause associated with each tree; and stores in the memory the one or more semantic structures in association with one or more connections between common ones of the tree elements, thereby generating a set of one or more semantic networks representing the resource. See also Dawson, [0089], where the system stores metadata in association with the one or more semantic structures, where metadata information includes storing as semantic relations at least one of: a numeric weight associated with at least one tree element, a comment associated with the at least one tree element, a comment associated with at least one connection, a tag for annotating the at least one tree element, an identifier for the tag, etc. (i.e., “along with multiple parameters relating to said resulting token”)) .
Although Dawson as modified does not appear to explicitly state that the “graph” is a “tree” as claimed, one of ordinary skill in the art would have recognized that the claimed “tree” is a type of “graph”, and one of ordinary skill in the art would have found it obvious to have modified Dawson to have utilized trees instead with the motivation of preserving structural ordering, as seen in human language, and as suggested by, e.g., Dawson, [FIG. 14], in the UI upper left panel regarding “special relativity”, with the child “is” and child “a physical theory”.
Dawson as modified does not appear to explicitly teach wherein for each sub-group of data that said collection of data is decomposed into, said sub-group is tokenized.
Fusco teaches wherein for each sub-group of data that said collection of data is decomposed into, said sub-group is tokenized (Fusco, [0041], where each multi-word expression (MWE) is encoded as a single token, as are resulting words which are also tokenized as single tokens. See Dawson, [0140-0156], with respect to “sub-group” resulting from the decomposition of the initial text).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Fusco (hereinafter “Dawson as modified”) by tokenizing multiple words with the motivation of enhancing semantic understanding by keeping certain words together, thereby potentially capturing a wider range of contexts and topics.
Although Dawson as modified does not appear to explicitly teach tokenizing “each” sub-group, e.g., Dawson’s paragraphs, sentences, etc., one of ordinary skill in the art would have found it obvious to have modified Dawson to do so with the motivation of capturing various sizes of meaning and contexts.
Regarding claim 14: Dawson teaches A system for implementing a language task …, the system comprising:
- a tokenizer for receiving and tokenizing a corpus of data to result in tokenized input data (Dawson, [0030], where information (in the form of semantic structures) is extracted from resources, which may comprise unstructured and structured data. See also Dawson, [0113], where resources are acquired via a discovery mechanism that utilizes semantic-based searching and web-crawling in the web. With resources in hand, the system then preprocesses text-based documents, specifically identifying text zones, splitting sentences and tokenizing, then performing a step of lexical and terminological analysis, and executing a step of syntactic analysis leading to incipient semantic structures, i.e., concept-relation-concept triples. These are ultimately used to produce one or more semantic structures representing a simple clause associated with each tree; and stores in the memory the one or more semantic structures in association with one or more connections between common ones of the tree elements, thereby generating a set of one or more semantic networks representing the resource, which includes storing one or more resource description framework (RDF) statements comprising RDF elements in association with one or more connections between common ones of the RDF elements. See also Dawson, [0256], where RDF graphs 518 that includes RDF elements, are stored in memory of the system, e.g., a database of the system);
wherein
- said tokenizer tokenizes at least a portion of said corpus of data by recursively decomposing said portion into smaller and smaller sub-groups of data, wherein … a resulting token is stored in a node in a tree along with multiple parameters relating to said resulting token … (Dawson, [0140-0156], where the system preprocesses documents to create a document container per document that describes structure and identifies parseable text zones; splits the text into paragraphs, paragraphs into sentences, and employs tokenizers to obtain lexical units (as seen, each of these groups becomes recursively decomposed into smaller and smaller units until tokens, or “lexical units” (i.e., “smallest units of data”) are obtained via tokenizers. These are ultimately used to produce one or more semantic structures representing a simple clause associated with each tree; and stores in the memory the one or more semantic structures in association with one or more connections between common ones of the tree elements, thereby generating a set of one or more semantic networks representing the resource).
Although Dawson as modified does not appear to explicitly state that the “graph” is a “tree” as claimed, one of ordinary skill in the art would have recognized that the claimed “tree” is a type of “graph”, and one of ordinary skill in the art would have found it obvious to have modified Dawson to have utilized trees instead with the motivation of preserving structural ordering, as seen in human language, and as suggested by, e.g., Dawson, [FIG. 14], in the UI upper left panel regarding “special relativity”, with the child “is” and child “a physical theory”.
Dawson does not appear to explicitly teach [implementing a language task] using an AI related model; a language task solver module receiving said tokenized input data and applying said Al related model to said tokenized input data; wherein for each sub-group of data that said portion is decomposed into, said sub-group is tokenized; [and] said language task solver is implemented in an execution block.
Sengupta teaches [implementing a language task] using an AI related model; a language task solver module receiving said tokenized input data and applying said Al related model to said tokenized input data (Sengupta, [0091-0096], where the significance recognition machine learning model 800 processes word-level representation data objects 802 and character-level representation data objects 804 via one or more attention stages, where multiple attention heads are specifically used to capture multiple different relationships between word-level tokens 602 (i.e., “tokenized input data”), etc., as well as generating a significance token label indicating a name significance subtype 604A for each word-level token); [and]
said language task solver is implemented in an execution block (Sengupta, [0006], where the disclosed significant recognition machine learning model may be implemented as a program product comprising program code portions that include execution portions configured to cause the system to implement the disclosed steps; see also, e.g., Sengupta, [0062], where the AI computing entity may retrieve and/or execute one or more of the predefined program algorithms upon the occurrence of a predefined trigger event).
Although Sengupta does not appear to explicitly state that the word tokens are “received” as claimed, one of ordinary skill in the art would have found it obvious to have modified Sengupta to receive the tokenized input data from, e.g., an upstream source, with the motivation of allowing the system to utilize preprocessed data, e.g., from an earlier time, rather than relying on having to process the source data itself, thereby conserving resources.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson and Sengupta (hereinafter “Dawson as modified”) with the motivation of enabling the system to learn, interpret, and understand human language in order to apply to a wide range of human-related tasks, e.g., healthcare services as described by Sengupta, or personal knowledge learning, as in Dawson, which enables greater relevancy and useability.
Dawson as modified does not appear to explicitly teach wherein for each sub-group of data that said portion is decomposed into, said sub-group is tokenized.
Fusco teaches wherein for each sub-group of data that said portion is decomposed into, said sub-group is tokenized (Fusco, [0041], where each multi-word expression (MWE) is encoded as a single token, as are resulting words which are also tokenized as single tokens. See Dawson, [0140-0156], with respect to “sub-group” resulting from the decomposition of the initial text).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Fusco (hereinafter “Dawson as modified”) by tokenizing multiple words with the motivation of enhancing semantic understanding by keeping certain words together, thereby potentially capturing a wider range of contexts and topics.
Although Dawson as modified does not appear to explicitly teach tokenizing “each” sub-group, e.g., Dawson’s paragraphs, sentences, etc., one of ordinary skill in the art would have found it obvious to have modified Dawson to do so with the motivation of capturing various sizes of meaning and contexts.
Regarding claim 17: Claim 17 recites substantially the same claim limitations as claim 13, and is rejected for the same reasons.
Claims 3-6 are rejected under 35 U.S.C. 103 as being unpatentable over Dawson et al. (“Dawson”) (US 2010/0228693 A1), in view of Sengupta et al. (“Sengupta”) (US 2023/0061731 A1), in further view of Fusco et al. (“Fusco”) (US 2023/0055769 A1), in further view of Loukas et al. (“Loukas”) (US 2023/0028664 A1).
Regarding claim 3: Dawson as modified teaches The method according to claim 2, wherein said preprocessing method includes repeating said decomposition step until said collection of data has been decomposed into smallest units of data … for said collection of data and until said smallest units of data have been tokenized and resulting tokens have been stored in said tree (Dawson, [0140-0156], where the system preprocesses documents to create a document container per document that describes structure and identifies parseable text zones; splits the text into paragraphs, paragraphs into sentences, and employs tokenizers to obtain lexical units (as seen, each of these groups becomes recursively decomposed into smaller and smaller units until tokens, or “lexical units” (i.e., “smallest units of data”) are obtained via tokenizers. These are ultimately used to produce one or more semantic structures representing a simple clause associated with each tree; and stores in the memory the one or more semantic structures in association with one or more connections between common ones of the tree elements, thereby generating a set of one or more semantic networks representing the resource) , and wherein said decomposition step includes grafting said tree resulting from said method to a larger tree (Dawson, [0273], where the system may connect any number of the semantic networks to any other number of semantic networks, thereby creating larger “islands” of semantic networks).
Although Dawson does not appear to explicitly state that the decomposition step includes connecting (i.e., “grafting”) the semantic network (i.e., corresponding to the claimed “tree”) to a “larger” semantic network (i.e., tree), the claimed invention would have been performed the same regardless of the size of the semantic network. Therefore, one of ordinary skill in the art would have found it obvious to have modified Dawson to have connected semantic networks to larger semantic networks with the motivation of building out more extensive semantic networks, i.e., creating larger “islands” of semantic networks (Dawson, [0273]), thereby expanding the semantic network having connections to new and existing concepts (see, e.g., Dawson, [0219]).
Dawson as modified does not appear to explicitly teach the data is decomposed into the smallest units of data “for a data type”.
Loukas teaches decomposing the data into the smallest unit of data “for a data type” (Loukas, [0049-0050], where tokenizing the extracted text into the plurality of tokens processes or handles the predetermined symbol as an undividable or unsplittable expression and/or transfers the predetermined symbol into a single token. See Dawson above with respect to decomposing into the “smallest unit”, i.e., words).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Loukas with the motivation of improving the performance of deep learning models by avoiding sub-word fragmentation of numbers and/or dates, which harms the performance of many deep learning models (Loukas, [0050]).
Regarding claim 4: Dawson as modified teaches The method according to claim 3, wherein said preprocessing method includes repeating said decomposition step until all of said corpus of data has been tokenized and resulting tokens have been grafted to said larger tree (Dawson, [0259], where the system can produce on RDF graph 522 for each natural language sentence 502 of the document 524, and repeat the steps 504, 512, 516 and 520 for each natural language sentence 502 of the submitted information. See also Dawson, [0074] and [0113], where repeated requests may be made to process documents and modify the corresponding semantic network).
Although Dawson does not appear to explicitly state that these steps are repeated with respect to processing the entire corpus of data, one of ordinary skill in the art would have found it obvious to have modified Dawson to apply these steps to an entire corpus with the motivation of limiting the number of resources to process, thereby conserving processing resources, and giving a substantial sample of information by which to build out knowledge, thereby potentially leading to greater accuracy in information.
Regarding claim 5: Dawson as modified teaches The method according to claim 4, wherein said larger tree is converted into said graph database (Dawson, [0088-0089], where generating a set of one or more semantic networks representing the resource, includes storing one or more resource description framework (RDF) statements comprising RDF elements in association with one or more connections between common ones of the RDF elements. See also Dawson, [0256], where RDF graphs 518 that includes RDF elements, are stored in memory of the system, e.g., a database of the system (implying that, for the storage to occur, a “convert[ing]” of the graph had occurred)).
Regarding claim 6: Dawson as modified teaches The method according to claim 4, wherein said graph database is a preexisting graph database and said larger tree is converted and incorporated into said preexisting graph database (Dawson, [0273], where the system may connect any number of the semantic networks to any other number of semantic networks, thereby creating larger “islands” of semantic networks (implying that the connected semantic networks are “preexisting”). See also Dawson, [0256], where RDF graphs 518 may be stored in memory of the system, e.g., in a database of the system (implying that the database is “preexisting” as claimed)).
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Dawson et al. (“Dawson”) (US 2010/0228693 A1), in view of Sengupta et al. (“Sengupta”) (US 2023/0061731 A1), in further view of Dimitriadis et al. (“Dimitriadis”) (US 20230297777 A1).
Regarding claim 7: Dawson as modified teaches The method according to claim 1, wherein said preprocessed data is tokenized data (Dawson, [0140-0156], [0211] and [FIG. 4, box 402], where the results of the natural language processing (or preprocessing performed on the documents) are used to decompose complex sentences into a set of simple clauses and lexical units or “tokens”) … .
Dawson as modified does not appear to explicitly teach step c includes retrieving multiple sets of tokenized data from said graph database, each of said multiple sets of tokenized data being subsets of said corpus of data.
Dimitriadis teaches step c includes retrieving multiple sets of tokenized data from said graph database, each of said multiple sets of tokenized data being subsets of said corpus of data (Dimitriadis, [0033-0034], where user specifier 34 receives a plurality of sets of tokenized text data 31, where each set of tokenized text data 31 includes a sequence of individual tokens 32 corresponding to the raw text 28, e.g., a plurality of sets of raw text data 28. Personalized NLP model 38 receives the user-specific token set 36 as input and processes the plurality of sets of tokenized text data 36 using the NLP model 38).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Dimitriadis (hereinafter “Dawson as modified”) with the motivation of having data in a standardized, small (unit) structure/format for NLP processing, which reduces computational complexity.
Claims 8-9 are rejected under 35 U.S.C. 103 as being unpatentable over Dawson et al. (“Dawson”) (US 2010/0228693 A1), in view of Sengupta et al. (“Sengupta”) (US 2023/0061731 A1), in further view of Dimitriadis et al. (“Dimitriadis”) (US 20230297777 A1), in further view of Shahar et al. (“Shahar”) (US 2023/0385541 A1).
Regarding claim 8: Dawson as modified teaches The method according to claim 7, but does not appear to explicitly teach wherein said execution block comprises an execution unit for applying an Al based model to a set of tokenized data and said multiple sets of tokenized data are sent to said execution unit in a sequential manner.
Shahar teaches wherein said execution block comprises an execution unit for applying an Al based model to a set of tokenized data and said multiple sets of tokenized data are sent to said execution unit in a sequential manner (Shahar, [0097], and [0157], and [0167], where the execution plan identifies one or more ML and/or NLP models required for execution of each of the selected NLP tasks, and NLP tasks may have a particular order and/or execution dependency between the selected NLP tasks (implying that the execution occurs sequentially), where NLP tasks my depend on, e.g., tokenization).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Shahar with the motivation of ensuring that (advanced) NLP capabilities, which commonly rely on other NLP capabilities to function, are executed in the proper order to ensure proper functioning (see, e.g., Shahar, [0023]).
Regarding claim 9: Dawson as modified teaches The method according to claim 7, but does not appear to explicitly teach wherein said execution block comprises a plurality of execution units operating in parallel, each of said plurality of execution units being for applying an Al based model to a set of tokenized data and said multiple sets of tokenized data are sent to said plurality of execution units in parallel such that said Al based model is applied to said sets of tokenized data simultaneously in parallel.
Shahar teaches wherein said execution block comprises a plurality of execution units operating in parallel, each of said plurality of execution units being for applying an Al based model to a set of tokenized data and said multiple sets of tokenized data are sent to said plurality of execution units in parallel such that said Al based model is applied to said sets of tokenized data simultaneously in parallel (Shahar, [0157] and [0166-0167], where the execution plan identifies one or more ML and/or NLP models required for execution of each of the selected NLP tasks, and an input can be provided simultaneously to different NLP tasks for parallel execution, where NLP tasks my depend on, e.g., tokenization).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Shahar with the motivation of speeding up processing by having different NLP models perform their respective NLP tasks.
Claims 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Dawson et al. (“Dawson”) (US 2010/0228693 A1), in view of Sengupta et al. (“Sengupta”) (US 2023/0061731 A1), in further view of Fusco et al. (“Fusco”) (US 2023/0055769 A1), in further view of Shahar et al. (“Shahar”) (US 2023/0385541 A1).
Regarding claim 15: Dawson as modified teaches The system according to claim 14, but does not appear to explicitly teach wherein said execution block comprises an execution unit implementing said Al related model on sequentially processed groups of tokenized input data.
Shahar teaches wherein said execution block comprises an execution unit implementing said Al related model on sequentially processed groups of tokenized input data (Shahar, [0097], and [0157], and [0167], where the execution plan identifies one or more ML and/or NLP models required for execution of each of the selected NLP tasks, and NLP tasks may have a particular order and/or execution dependency between the selected NLP tasks (implying that the execution occurs sequentially), where NLP tasks my depend on, e.g., tokenization).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Dawson as modified and Shahar with the motivation of ensuring that (advanced) NLP capabilities, which commonly rely on other NLP capabilities to function, are executed in the proper order to ensure proper functioning (see, e.g., Shahar, [0023]).
Regarding claim 16: Claim 16 recites substantially the same claim limitations as claim 9, and is rejected for the same reasons.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure. See the enclosed 892 form. Yan et al. US 2022/0138262 A1 is cited to show why one of ordinary skill in the art would have found it obvious to have utilized a graph database for storing data (Yan et al., [0002]). The prior art should be considered to define the claims over the art of record.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IRENE BAKER whose telephone number is (408)918-7601. The examiner can normally be reached M-F 8-5PM PT.
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, BORIS GORNEY can be reached at (571)270-5626. 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.
/IRENE BAKER/Primary Examiner, Art Unit 2154
16 April 2026
1 Yan et al. US 2022/0138262 A1 at [0002] (“Current trends include utilizing a graph model structure or a graph database to store data that is highly connected to provide flexibility in adding data, running faster relationship-based searches, and indexing by relationships….”).
Prosecution Insights