DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This Action is non-final and is in response to the claims filed February 5, 2026 via RCE. Claims 2-25 are currently pending, of which claims 2, 9, and 16 are currently amended. Claims 23-25 are newly presented and claim 1 was previously cancelled.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on February 5, 2026 has been entered.
Response to Arguments
Double Patenting
In accordance with Applicant’s request, “the nonstatutory obviousness-type double patenting rejection will be held in abeyance until such time as allowable subject matter is identified for the present Application.” See Remarks 6. Examiner notes that in light of the claim amendments, the rejection has been updated below.
Prior Art Rejections
Applicant’s arguments regarding the previously cited art have been fully considered and are not persuasive.
Specifically, Applicant argues that Liu does not add “texts to create an augmented text that is used to generate a response from a machine learning model” and that the claimed augmented text is not in vector format. Moreover, Applicant argues that “the one or more semantically related texts are added to the first text, not to the vector representation of the first text.” See Remarks 8. Examiner respectfully disagrees with both Applicant’s arguments and characterization of the prior art references.
Firstly, Liu explicitly teaches vector representations of text. See Liu Fig. 3 and paras. [0023] and [0035]. More specifically, these vectors Liu then explicitly uses these vector representations in an aggregate (i.e. added) form that can then be part of the vector representations of that first text. See Id. at Fig. 3 and para. [0039]. This is not just related texts being added to the first text but instead are vector representations.
Nevertheless, Subasic also discloses these features of the claim and the rejection has been updated below accordingly. Specifically, Subasic discloses combining text corpuses into a combined vector model from pre-trained vectors (whether that corpus is a public word2vec model training or if it is new model training). Those models are then combined. See Subasic Fig. 2 and paras. [0012-15]. However, as discussed above, these features are taught by Liu.
Finally, again assuming that Applicant’s arguments were convincing, this completely leaves out what could be obvious to one of ordinary skill in the art at the time the invention was filed. Specifically, substituting the location of where the text is added (the vector representation versus the first text) would arguably be an obvious variant. Nevertheless, this conclusion isn’t necessary because, as discussed above, Liu (and Subasic) does teach the language at issue.
Examiner encourages language that further clarifies the anomaly detection, what an anomaly actually is, and how those anomalies are used and intertwined with the texts and vectors of the current independent claims.
It is for at least these reasons, and the reasons cited below, that the claims remain rejected in this Action.
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 2-25 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1-20 of U.S. Patent No. 12,039,276 B2 (hereinafter, “the ‘276 patent”) in view of Liu et al. (U.S. Publication No. 2020/0372115; hereinafter, “Liu”).
As per claim 2, claim 1 of the ‘276 patent discloses the majority of the claim language. However, the ‘276 patent does not disclose vector representations. Liu teaches the vector representations as well as generating a response from a machine learning model (See Liu para. [0023]: “words or phrases from the vocabulary are mapped to vectors of real numbers. Ideally, an embedding places semantically similar inputs close together in the embedding space to capture the semantics of the inputs”; paras. [0055-56]: output embeddings classified together in a shared vector space).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine, with a reasonable expectation of success, the text classifications and groups of the ‘276 patent with the vectors and machine learning model outputs of Liu. One would have been motivated to combine these references because both references disclose natural language processing as it relates to semantically classifying and aggregating text in domain-specific manners. Liu enhances the categorization of the ‘276 patent by improving the functionality and operation of a cognitive systems by efficiently providing for language and domain acceleration with embedding evaluation for improved generation of natural language understanding models (See Liu paras. [0011], [0021], and [0023]).
As per claim 3, the ‘276 patent does not disclose the training data set for the machine learning model.
Liu teaches the method of claim 2, wherein the one or more domain-specific texts were not part of a training data set for the machine learning model (See Liu para. [0024]: “the cross-lingual learning module 16 uses any suitable technique to process the trained monolingual embeddings 22 in different languages so that they are aligned in a shared space where words of high semantic similarity across languages are close to each other. The aligned embeddings are referred to as multilingual embeddings. As will be appreciated by those skilled in the art, cross-lingual learning may be implemented by constructing a parallel vocabulary from key or “anchor” words (e.g., frequent unigrams) in each monolingual embedding, and then using the parallel vocabulary as anchor points to transform a first or “source” embedding space into a second or “target” embedding space.” This is not the embedded training data, as it occurs after training the model)
Claim 4 of the present invention is taught by at least claim 2 of the ‘276 patent.
Claim 5 of the present invention is not taught by the ‘276 patent. However, Liu teaches the method of claim 2, wherein comparing the vector representation of the first text to a domain-specific data set is performed using natural language processing techniques (See Liu para. [0035]: “the language identification module/process 314 may extract and analyze words from the text data 312 for comparison to dictionaries from different languages in order to identify the language for each text data file 312”).
Claim 6 of the present invention is taught by at least claim 6 of the ‘276 patent. However, the ‘276 patent does not teach the cosine similarity measurement associated with the plurality of vectors.
Liu further teaches the method of claim 2, wherein comparing the vector representation of the first text to a domain-specific data set is performed using a cosine similarity measurement between the vector representation of the first text and the plurality of vectors generated by a machine learning process, each vector representing a domain-specific text (See Liu paras. [0026-27] and [0038-39]: “embedding evaluation may be implemented by determining the association between two given words using a calculation of the cosine similarity between the embedding vectors for the words.” Furthermore, “the multi-lingual/domain embedding engine 14 may include a machine learning model generator 19 for processing the cross-domain, multilingual embeddings 23 into one or more language-independent and domain-independent natural language models 24…”).
Claims 7 and 8 of the present invention is taught by at least claim 1 of the ‘276 patent.
Claim 23 of the present invention is taught by at least claim 1 of the ‘276 patent. Specifically, there is anomalous text associated that would therefore be related to an anomaly as claimed in the present application.
As per claims 9-15 and 24, the claims are directed to a system that implements the same features as the method of claims 2-8 and 23, respectively, and are therefore rejected for at least the same reasons therein. This includes the processors and memory of claim 10 of the ‘276 patent.
As per claims 16-22 and 25, the claims are directed to a computer-readable medium that implements the same features as the method of claims 2-8 and 23, respectively, and are therefore rejected for at least the same reasons therein. This includes the storage media of claim 18 of the ‘276 patent.
Examiner’s Note
The prior art rejections below cite particular paragraphs, columns, and/or line numbers in the references for the convenience of the applicant. Although the specified citations are representative of the teachings in the art and are applied to the specific limitations within the individual claim, other passages and figures may apply as well. It is respectfully requested that, in preparing responses, the applicant fully consider the references in their entirety as potentially teaching all or part of the claimed invention, as well as the context of the passage as taught by the prior art.
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.
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.
Claim(s) 2-22 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu et al. (U.S. Publication No. 2020/0372115; hereinafter, “Liu”) and further in view of Subasic et al. (U.S. Publication No. 2019/0243904; hereinafter “Subasic”).
As per claim 2, Liu teaches a method comprising:
receiving a first text (See Liu paras. [0021-22]: receive and process input text data);
identifying one or more semantically related texts by comparing a vector representation of the first text to a domain-specific data set, the domain-specific data set comprising a plurality of vectors [previously generated by a machine learning process], each vector of the domain-specific data set representing one or more domain-specific texts (See Liu Fig. 3 and paras. [0023] and [0035]: “words or phrases from the vocabulary are mapped to vectors of real numbers. Ideally, an embedding places semantically similar inputs close together in the embedding space to capture the semantics of the inputs”. Furthermore, comparisons can be made against different domains and the training module “may use any suitable vectorization mechanism to process the text data 312 to generate vectors to represent words to provide a distributed representation of the words in a language”);
creating an augmented text by adding the first text to the one or more semantically related texts (See Liu paras. [0039-40]: cross-domain alignment module may combine embodiments to generate an aggregated embedding space from the two corresponding embedding spaces. “Given two sets of target words and two sets of attribute words from two different domains, if the two monolingual embeddings are aligned perfectly, there should be no difference between the target words in terms of their relative similarity to the attribute words”); and
using the augmented text to generate a response from a machine learning model (See Liu paras. [0055-56]: output embeddings classified together in a shared vector space).
However, while Liu discloses machine learning processes, Liu does not explicitly teach or suggest that the vectors are previously generated by that machine learning process.
Subasic teaches that the vectors of Liu can be previously generated by a machine learning process (See Subasic Fig. 2 and paras. [0012-15]: combining text corpuses into a combined vector model from pre-trained vectors (whether that corpus is a public word2vec model training or if it is new model training). Those models are then combined).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine, with a reasonable expectation of success, the vector mappings of Liu with the pre-trained vector model of Subasic. One would have been motivated to combine these references because both references disclose vector mapping for text augmentation/embedding, and Subasic enhances the modeling of Liu by obviating “cumbersome training and enables the enhancement of a pre-existing vector model through a new corpus corresponding to a new vector model by merely transforming the vectors from the pre-existing vector model and from the new vector model into the vectors for the combined vector model” (See Subasic para. [0007]).
As per claim 3, Liu/Subasic further teaches the method of claim 2, wherein the one or more domain-specific texts were not part of a training data set for the machine learning model (See Liu para. [0024]: “the cross-lingual learning module 16 uses any suitable technique to process the trained monolingual embeddings 22 in different languages so that they are aligned in a shared space where words of high semantic similarity across languages are close to each other. The aligned embeddings are referred to as multilingual embeddings. As will be appreciated by those skilled in the art, cross-lingual learning may be implemented by constructing a parallel vocabulary from key or “anchor” words (e.g., frequent unigrams) in each monolingual embedding, and then using the parallel vocabulary as anchor points to transform a first or “source” embedding space into a second or “target” embedding space.” This is not the embedded training data, as it occurs after training the model).
As per claim 4, Liu/Subasic further teaches the method of claim 2, wherein the first text is unstructured text (See Liu para. [0021]: text data stored in the knowledge base (and used for input) can be unstructured content).
As per claim 5, Liu/Subasic further teaches the method of claim 2, wherein comparing the vector representation of the first text to a domain-specific data set is performed using natural language processing techniques (See Liu para. [0035]: “the language identification module/process 314 may extract and analyze words from the text data 312 for comparison to dictionaries from different languages in order to identify the language for each text data file 312”).
As per claim 6, Liu/Subasic further teaches the method of claim 2, wherein comparing the vector representation of the first text to a domain-specific data set is performed using a cosine similarity measurement between the vector representation of the first text and the plurality of vectors generated by a machine learning process, each vector representing a domain-specific text (See Liu paras. [0026-27] and [0038-39]: “embedding evaluation may be implemented by determining the association between two given words using a calculation of the cosine similarity between the embedding vectors for the words.” Furthermore, “the multi-lingual/domain embedding engine 14 may include a machine learning model generator 19 for processing the cross-domain, multilingual embeddings 23 into one or more language-independent and domain-independent natural language models 24…”).
As per claim 7, Liu/Subasic further teaches the method of claim 2, wherein the response includes a classification of information from the first text (See Liu paras. [0055-56]: output embeddings classified together in a shared vector space. This includes similar language words such as Italian, German, and/or French).
As per claim 8, Liu/Subasic further teaches the method of claim 2, wherein the first text comprises a word list (See Liu para. [0039]: collections of stopwords).
As per claims 9-15 and 24, the claims are directed to a system that implements the same features as the method of claims 3-8 and 23, respectively, and are therefore rejected for at least the same reasons therein. Furthermore, Liu teaches one or more devices, each device including one or more processors and a memory, wherein the system is configured to receive a series of instructions, which when executed on the one or more processors across the one or more devices, cause the system to perform actions including said method(s) (See Liu para. [0028]).
As per claims 16-22 and 25, the claims are directed to a computer-readable medium that implements the same features as the method of claims 3-8 and 23, respectively, and are therefore rejected for at least the same reasons therein. Furthermore, Liu teaches a non-transitory computer-readable medium, the medium including instructions which, when executed on one or more processors across one or more devices, cause the one or more devices to perform actions including said method(s) (See Liu para. [0014]).
Claims 23-25 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu/Subasic as applied above, and further in view of Ho et al. (U.S. Publication No. 2017/0132288; hereinafter, “Ho”).
As per claim 23, while Liu/Subasic teaches further teaches the method of claim 2, Liu/Subasic does not teach anomaly detection.
Ho teaches wherein the first text relates to an anomaly (See Ho paras. [0051-52] and [0082]: anomaly detection using vector representations, including identifying possible errors).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to combine, with a reasonable expectation of success, the vectors and text of Liu/Subasic with the anomaly detection of Ho. One would have been motivated to combine these references because both references disclose vector representations of concepts, and Ho enhances the modeling of Liu/Subasic by improving the effectiveness of the vector comparisons and allowing more efficient noise reductions while also optimizing a desired goal (See Ho paras. [0002] and [0047]).
As per claim 24, the claim is directed to a system that implements the same features as the method of claim 23 and is therefore rejected for at least the same reasons therein.
As per claim 25, the claim is directed to a computer-readable medium that implements the same features as the method of claim 23 and is therefore rejected for at least the same reasons therein.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Nicholas Klicos whose telephone number is (571)270-5889. The examiner can normally be reached Mon-Fri 9:00 AM-5:00 PM.
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, Scott Baderman can be reached at (571) 272-3644. 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.
/NICHOLAS KLICOS/Primary Examiner, Art Unit 2118