DETAILED ACTION
Introduction
This office action is in response to Applicant’s amendment filed on March 2, 2026.
Claims 1, 4, 6, 12 and 16-18 have been amended. Claims 3 and 10 have been cancelled. New claims 19-20 have been added. Claims 1-2, 4-9 and 11-20 are pending in the application. As such, claims 1-2, 4-9 and 11-20 have been examined.
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 .
Drawings
The drawings were received on March 18, 2025. These drawings have been accepted and considered by the Examiner.
Specification
Applicant is reminded of the proper language and format for an abstract of the disclosure.
The abstract should be in narrative form and generally limited to a single paragraph on a separate sheet within the range of 50 to 150 words in length. The abstract should describe the disclosure sufficiently to assist readers in deciding whether there is a need for consulting the full patent text for details.
The language should be clear and concise and should not repeat information given in the title. It should avoid using phrases which can be implied, such as, “The disclosure concerns,” “The disclosure defined by this invention,” “The disclosure describes,” etc. In addition, the form and legal phraseology often used in patent claims, such as “means” and “said,” should be avoided.
Appropriate correction is required.
Response to Amendments and Arguments
In view of the amendments to the claims, the amendments to the claims have been acknowledged and entered.
In view of the amendments to the claims, the objections to claims 3, 4, 6 and 12, have been withdrawn.
In view of the amendments to the claims, the rejections to claims 1-2, 10, 12 and 16-17, under 35 U.S.C. 101 have been withdrawn.
In view of the amendments to the claims, the rejections to claims 1-18 under 35 U.S.C. 103 have been withdrawn.
In light of the amendments to the claims, new grounds for rejection for claims 1-2, 4-9 and 11-20, under 35 U.S.C. 103 are provided in the response below. New grounds for rejection is based at least upon the following new elements:
integrating an inmate communications language model with at least one inmate communications search application via a model application programming interface; [[and]]
performing word embeddings to map a plurality of inmate communications tokens derived from a plurality of inmate communications transcripts to a corresponding plurality of inmate communications vectors; and
training the inmate communications language model based on a subset of the plurality of inmate communications vectors;
providing selectable real-time lexicographic recommendations in response to at least one inmate communications search query, the selectable real-time lexicographic recommendations comprising at least one of a synonym, slang term, mistranscription correction, and misspelling correction[[.]]; and
receiving recommendation feedback comprising lexicographic recommendation selections in response to the selectable real-time lexicographic recommendations.
Applicant’s arguments regarding the prior art rejections under 35 U.S.C 103, received on March 2, 2026, have been fully considered.
Applicant argues A – on page 5 (Request 1), “Applicant respectfully requests clarification as to exactly where in the cited art the Examiner believes "selectable real-time lexicographic recommendations ... comprising at least one of a synonym, slang term, mistranscription correction, and misspelling correction," and, "recommendation feedback comprising lexicographic recommendation selections," is believed to be shown (REQUEST 1).”
Examiner response A - applicant’s argument is directed to the newly amended matter in the claims, and is addressed accordingly in the updated rejection rationale below.
Applicant argues B – on page 6 (Request 2), “Applicant respectfully requests that the Examiner provide a clear articulation of the basis for any allegation that there would be a reasonable expectation of success in making each of the proposed modifications and combinations of the cited references (REQUEST 2).”
Examiner response B – In response to applicant's argument, the test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981). In this case, the examiner would expect a working feature from one reference to work alongside another feature from another reference, unless there is evidence to the contrary in either reference.
Applicant argues C – on page 6 (Request 3), “Applicant respectfully requests clarification as to why the Examiner believes the "facts gleaned from the prior art" themselves, and not drawn in hindsight from "the claimed invention" justify the allegation of obviousness (REQUEST 3).”
Examiner response C – In response to applicant's argument that the examiner's conclusion of obviousness is based upon improper hindsight reasoning, it must be recognized that any judgment on obviousness is in a sense necessarily a reconstruction based upon hindsight reasoning. But so long as it takes into account only knowledge which was within the level of ordinary skill at the time the claimed invention was made, and does not include knowledge gleaned only from the applicant's disclosure, such a reconstruction is proper. See In re McLaughlin, 443 F.2d 1392, 170 USPQ 209 (CCPA 1971). In this case, the examiner took into account only knowledge which was within the level of ordinary skill at the time the claimed invention was made, and does not include knowledge gleaned only from the applicant's disclosure
Applicant argues D – on page 7 (Request 4), “Applicant respectfully requests clarification as to why a person of ordinary skill would be motivated to combine the references based on facts drawn from the prior art showing connections between and in relation to "the subject matter as a whole" (REQUEST 4).”
Examiner response D – In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, the motivations to combine the references remain the same as previously presented, and are provided in the updated 103 rejection below.
Applicant argues E – on page 8 (Request 5), “Applicant respectfully requests specific discussion of the alleged level of ordinary skill in the art at the time of the invention and clarification as to how and specifically why the Examiner believes the claimed subject matter would have been obvious to one possessing that level of skill (REQUEST 5).”
Examiner response E – In response to applicant’s argument that there is no teaching, suggestion, or motivation to combine the references, the examiner recognizes that obviousness may be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so found either in the references themselves or in the knowledge generally available to one of ordinary skill in the art. See In re Fine, 837 F.2d 1071, 5 USPQ2d 1596 (Fed. Cir. 1988), In re Jones, 958 F.2d 347, 21 USPQ2d 1941 (Fed. Cir. 1992), and KSR International Co. v. Teleflex, Inc., 550 U.S. 398, 82 USPQ2d 1385 (2007). In this case, the prior art identifies the available level of ordinary skill in the art at the time of the invention, and the claimed subject matter would have been obvious to one possessing that level of skill since the published documents are available for one of ordinary skill to access.
Applicant’s remaining arguments with respect to claims 1-2, 4-9 and 11-20 have been considered, are directed to the newly amended matter in the claims, are not considered to be persuasive, and are addressed accordingly in the updated rejection rationale below.
Claim Objections
Claim 12 is objected to because of the following informalities:
Claim 12, line 4, reads “transcribing inmate communications data”. Examiner believes this to be a clerical error and it is intended to read “transcribing the inmate communications data”.
Appropriate correction is required.
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.
Claims 1-2, 4-9, 11-17 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Broidy et al. (US Patent Pub. No. 20200366786 A1), hereinafter Broidy, in view of Hendry (US Patent Pub. No. 20190147000 A1), in view of Lu et al. (US Patent Pub. No. 20250173782 A1), hereinafter Lu, in view of Gaudiano et al. (US Patent Pub. No. 20120084149 A1), hereinafter Gaudiano.
Regarding claim 1, Broidy teaches a processor-implemented method (Broidy in [0031] teaches a method for processing audio data and indexing the audio data in a searchable database, and in [0066] teaches using processors for doing the processing),
comprising:
at least one inmate communications search application (Broidy in [0025] teaches calls are transcribed shortly after being completed between callers and callees (or multiple parties), and the transcription and other metadata regarding the calls are stored in a database in a searchable and indexed form, and users can perform searches for callers, callees, keywords, and/or other information in calls across the system, and in [0024] teaches making the resulting data searchable and accessible via a user interface);
a plurality of inmate communications transcripts (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like);
and
providing [selectable] real-time lexicographic recommendations in response to at least one inmate communications search query (Broidy in [0045-0047] teaches using a system which can search inmate call data and provide various ways to view the lexicon content, such as views at various granularities (year, month, week, day, hour, day of week, hour of day) to identify any patterns of behavior of the individuals or equipment),
the [selectable] real-time lexicographic recommendations comprising at least one of a synonym, slang term, mistranscription correction, and misspelling correction (Broidy in [0047] teaches using a system which can manage lexicon content (words, synonyms, aggregation hierarchy)).
Broidy does not teach, however Hendry teaches
an inmate communications language model (Hendry in [0047-0050] teaches that many language models are available, and that particular language models suit a particular purpose better or worse than other language models depending on circumstances relating to the particular purpose, and a language model in the context of the present application determines how relevant a search query is for a particular document based on a frequency with which each term is present in the document and the frequency with which that term is present in other documents within the corpus, and applying a language model, the M most relevant search queries for a document can be identified);
a [plurality of inmate communications] tokens [derived from a plurality of inmate communications transcripts] (Hendry in [0034] teaches parsing documents into tokens).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy further in view of Hendry to allow for selecting a model. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Broidy, as modified above, teaches the at least one inmate communications search application, and the inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
integrating the [inmate communications language model] with [at least one inmate communications search application] via a model application programming interface (Lu in [0052] teaches using an API to integrate a server with a device);
performing word embeddings to map a [plurality of inmate communications tokens derived from a plurality of inmate communications transcripts] to a corresponding [plurality of inmate communication] vectors (Lu in [0025] teaches generating embeddings corresponding to vector representations from the data);
and
training the [inmate communications model] based on a subset of the [plurality of inmate communication] vectors (Lu in [0048] teaches training data is used to train the machine learning model, and [0025] teaches converting the data into vectors);
receiving recommendation feedback comprising lexicographic recommendation selections in response to the [selectable real-time lexicographic recommendations] (Lu in [0081] teaches the user may provide feedback on one or more of the predictions received).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for using an API. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Broidy, as modified above, does not teach, however Gaudiano teaches
selectable real-time lexicographic recommendations (Gaudiano in [0020] teaches a user interacting with a broader number of keywords presented in such interactive text clouds is more likely to find and select keywords that are of particular interest).
Gaudiano is considered to be analogous to the claimed invention because it is in the same field of using interactive text clouds. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Gaudiano to allow for using interactive text clouds. Motivation to do so would allow for the user to interact with the words, e.g., by selecting some words or topics (for instance by clicking on them), removing some unwanted words or topics, focusing on specific words or topics, changing the number of words or topics displayed, or altering the way in which words or topics are displayed (Gaudiano [0012]).
Regarding claim 2, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, does not teach, however Hendry teaches
further comprising:
selecting a language model architecture for an inmate communications language model (Hendry in [0047-0050] teaches that many language models are available, and that particular language models suit a particular purpose better or worse than other language models depending on circumstances relating to the particular purpose, and a language model in the context of the present application determines how relevant a search query is for a particular document based on a frequency with which each term is present in the document and the frequency with which that term is present in other documents within the corpus, and applying a language model, the M most relevant search queries for a document can be identified).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Hendry to allow for selecting a model. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Regarding claim 4, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, teaches the plurality of inmate communications.
Broidy, as modified above, does not teach, however Lu teaches
further comprising:
splitting the plurality of inmate communications vectors into training vectors and test vectors (Lu in [0048] teaches validation and testing are performed for a machine learning model, such as based on validation data and test data, as is known in the art, and [0025] teaches converting the [two types of] data into vectors);
and
wherein the subset of the plurality of inmate communications vectors correspond to the training vectors (Lu in [0048] teaches training data is used to train the machine learning model, and [0025] teaches converting the data into vectors).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for training data to be used to train a machine learning model. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Regarding claim 5, Broidy, as modified above, teaches the method of claim 4.
Broidy, as modified above, teaches the inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
further comprising:
validating the [inmate communications language model] with the testing vectors (Lu in [0048] teaches validation data is used to validate the machine learning model, and [0025] teaches converting the data into vectors);
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for validating data to be used to validate a machine learning model. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Regarding claim 6, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, teaches the plurality of inmate communications and the plurality of inmate communication transcripts.
Broidy does not teach, however Hendry teaches
further comprising:
tokenizing the [plurality of inmate communications transcripts] to yield the [plurality of inmate communications] tokens (Hendry in [0034] teaches parsing documents into tokens).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Hendry to allow for parsing documents into tokens. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Regarding claim 7, Broidy, as modified above, teaches the method of claim 6.
Broidy, as modified above, teaches the plurality of inmate communications.
Broidy does not teach, however Hendry teaches
further comprising:
removing stop words from the [plurality of inmate communications] tokens (Hendry in [0034] teaches a “stop list” may also be utilized when parsing a document, where a stop list may direct the parser to ignore certain tokens that would not be useful to forming relevant search queries from the selected document, and grammatical articles such as “a”, “an”, and “the” are examples of tokens frequently in stop lists);
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Hendry to allow for ignoring stop words. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Regarding claim 8, Broidy, as modified above, teaches the method of claim 6.
Broidy further teaches
further comprising:
transcribing a plurality of inmate communications data to yield the plurality of inmate communications transcripts (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like).
Regarding claim 9, Broidy, as modified above, teaches the method of claim 8.
Broidy further teaches
further comprising:
receiving the inmate communications data (Broidy in [0056-0057] teaches receiving call data when monitoring conversations by inmates).
Regarding claim 11, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, teaches the inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
further comprising:
updating at least one [inmate communications language model] parameter based on the recommendation feedback (Lu in [0081] teaches the labeled training data for the machine learning model may be updated based on the feedback and the updated labeled training data may be used to re-train the machine learning model).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for re-training a machine learning model. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Regarding claim 12, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, teaches the inmate communications language model.
Broidy further teaches
a plurality of inmate communications transcripts (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like).
Broidy, as modified above, teaches the plurality of inmate communications, and the inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
a corresponding [plurality of inmate communication] vectors (Lu in [0025] teaches generating embeddings corresponding to vector representations from the data).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for generating embeddings corresponding to vector representations. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Broidy, as modified above, teaches the plurality of inmate communications, the vectors, and the inmate communications language model.
Broidy, as modified above, does not teach, however Gaudiano teaches
further comprising:
providing an interactive word cloud via a user interface based on the [inmate communications language model] (Gaudiano in [0053] teaches an example interactive text cloud displayed by a widget, and the example interactive text cloud includes twenty four words or topics which have been extracted from a variety of source text(s) in a format known as RSS feed, where an RSS feed comprises a text file containing some headings and descriptive labels to facilitate display of the text for an end user),
wherein proximity of words in the interactive word cloud is based on a relationship between a corresponding [plurality of inmate communications vectors in the inmate communications language model] (Gaudiano in [0053] teaches the size of each word or topics is indicative of its frequency relative to the frequency of other words or topics, and the location of words or topics relative to one another, as well as the thin lines connecting some words or topics, is indicative of the fact that certain words or topics tend to occur in close proximity to one another in the source text(s)).
Gaudiano is considered to be analogous to the claimed invention because it is in the same field of using interactive text clouds. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Gaudiano to allow for using interactive text clouds. Motivation to do so would allow for the user to interact with the words, e.g., by selecting some words or topics (for instance by clicking on them), removing some unwanted words or topics, focusing on specific words or topics, changing the number of words or topics displayed, or altering the way in which words or topics are displayed (Gaudiano [0012]).
Regarding claim 13, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, teaches the inmate communications language model.
Broidy further teaches
further comprising:
transcribed inmate communications (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like)
applying the [inmate communications language model] to newly transcribed inmate communications to identify novel terms (Broidy in [0072] teaches the system locates an existing lexicon node with the word and/or as a synonym for that node, and if no such node is found, the word is recorded as a new word and the call id and caller id are associated with that entry, which allows analysts to review new words for inclusion in the lexicon, and to access their use in context of calls and the callers making such calls);
updating the [inmate communications language model] based on the novel terms (Broidy in [0072] teaches the system locates an existing lexicon node with the word and/or as a synonym for that node, and if no such node is found, the word is recorded as a new word and the call id and caller id are associated with that entry, which allows analysts to review new words for inclusion in the lexicon, and to access their use in context of calls and the callers making such calls).
Regarding claim 14, Broidy, as modified above, teaches the method of claim 13.
Broidy further teaches
further comprising:
transcribing inmate communications data to yield the newly transcribed inmate communications (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like).
Regarding claim 15, Broidy, as modified above, teaches the method of claim 14.
Broidy further teaches
further comprising:
receiving the inmate communications data (Broidy in [0056-0057] teaches receiving call data when monitoring conversations by inmates).
Regarding claim 16, Broidy teaches an apparatus (Broidy in [0024] teaches an apparatus that obtains audio and metadata information from voice calls, generates textual transcripts from those calls, and makes the resulting data searchable and accessible via a user interface),
comprising:
a processor; a memory communicatively coupled to the processor and storing program instructions that, when executed, cause the processor to (Broidy in [0066] teaches using processors for doing the processing, and which are able to execute instructions stored in a memory):
at least one inmate communications search application (Broidy in [0025] teaches calls are transcribed shortly after being completed between callers and callees (or multiple parties), and the transcription and other metadata regarding the calls are stored in a database in a searchable and indexed form, and users can perform searches for callers, callees, keywords, and/or other information in calls across the system, and in [0024] teaches making the resulting data searchable and accessible via a user interface);
and
provide [selectable] real-time lexicographic recommendations in response to at least one inmate communications search query (Broidy in [0045-0047] teaches using a system which can search inmate call data and provide various ways to view the lexicon content, such as views at various granularities (year, month, week, day, hour, day of week, hour of day) to identify any patterns of behavior of the individuals or equipment),
the [selectable] real-time lexicographic recommendations comprising at least one of a synonym, slang term, mistranscription correction, and misspelling correction (Broidy in [0047] teaches using a system which can manage lexicon content (words, synonyms, aggregation hierarchy)).
Broidy does not teach, however Hendry teaches
an inmate communications language model (Hendry in [0047-0050] teaches that many language models are available, and that particular language models suit a particular purpose better or worse than other language models depending on circumstances relating to the particular purpose, and a language model in the context of the present application determines how relevant a search query is for a particular document based on a frequency with which each term is present in the document and the frequency with which that term is present in other documents within the corpus, and applying a language model, the M most relevant search queries for a document can be identified);
a [plurality of inmate communications] tokens [derived from a plurality of inmate communications transcripts] (Hendry in [0034] teaches parsing documents into tokens).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy further in view of Hendry to allow for selecting a model. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Broidy, as modified above, teaches the at least one inmate communications search application, and inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
integrate an [inmate communications language model] with [at least one inmate communications search application] via a model application programming interface (Lu in [0052] teaches using an API to integrate a server with a device);
performing word embeddings to map a [plurality of inmate communications tokens derived from a plurality of inmate communications transcripts] to a corresponding [plurality of inmate communication] vectors (Lu in [0025] teaches generating embeddings corresponding to vector representations from the data);
and
training the [inmate communications model] based on a subset of the [plurality of inmate communication] vectors (Lu in [0048] teaches training data is used to train the machine learning model, and [0025] teaches converting the data into vectors);
receiving recommendation feedback comprising lexicographic recommendation selections in response to the [selectable real-time lexicographic recommendations] (Lu in [0081] teaches the user may provide feedback on one or more of the predictions received).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for using an API. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Broidy, as modified above, does not teach, however Gaudiano teaches
selectable real-time lexicographic recommendations (Gaudiano in [0020] teaches a user interacting with a broader number of keywords presented in such interactive text clouds is more likely to find and select keywords that are of particular interest).
Gaudiano is considered to be analogous to the claimed invention because it is in the same field of using interactive text clouds. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Gaudiano to allow for using interactive text clouds. Motivation to do so would allow for the user to interact with the words, e.g., by selecting some words or topics (for instance by clicking on them), removing some unwanted words or topics, focusing on specific words or topics, changing the number of words or topics displayed, or altering the way in which words or topics are displayed (Gaudiano [0012]).
Regarding claim 17, Broidy teaches a processor-accessible, non-transitory medium storing processor-issuable program instructions (Broidy in [0066] teaches using processors for doing the processing, and which are able to execute instructions stored in a memory),
comprising:
at least one inmate communications search application (Broidy in [0025] teaches calls are transcribed shortly after being completed between callers and callees (or multiple parties), and the transcription and other metadata regarding the calls are stored in a database in a searchable and indexed form, and users can perform searches for callers, callees, keywords, and/or other information in calls across the system, and in [0024] teaches making the resulting data searchable and accessible via a user interface);
and
provide [selectable] real-time lexicographic recommendations in response to at least one inmate communications search query (Broidy in [0045-0047] teaches using a system which can search inmate call data and provide various ways to view the lexicon content, such as views at various granularities (year, month, week, day, hour, day of week, hour of day) to identify any patterns of behavior of the individuals or equipment),
the [selectable] real-time lexicographic recommendations comprising at least one of a synonym, slang term, mistranscription correction, and misspelling correction (Broidy in [0047] teaches using a system which can manage lexicon content (words, synonyms, aggregation hierarchy)).
Broidy does not teach, however Hendry teaches
an inmate communications language model (Hendry in [0047-0050] teaches that many language models are available, and that particular language models suit a particular purpose better or worse than other language models depending on circumstances relating to the particular purpose, and a language model in the context of the present application determines how relevant a search query is for a particular document based on a frequency with which each term is present in the document and the frequency with which that term is present in other documents within the corpus, and applying a language model, the M most relevant search queries for a document can be identified);
a [plurality of inmate communications] tokens [derived from a plurality of inmate communications transcripts] (Hendry in [0034] teaches parsing documents into tokens).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy further in view of Hendry to allow for selecting a model. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Broidy, as modified above, teaches the at least one inmate communications search application, and inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
integrate an [inmate communications language model] with [at least one inmate communications search application] via a model application programming interface (Lu in [0052] teaches using an API to integrate a server with a device);
performing word embeddings to map a [plurality of inmate communications tokens derived from a plurality of inmate communications transcripts] to a corresponding [plurality of inmate communication] vectors (Lu in [0025] teaches generating embeddings corresponding to vector representations from the data);
and
training the [inmate communications model] based on a subset of the [plurality of inmate communication] vectors (Lu in [0048] teaches training data is used to train the machine learning model, and [0025] teaches converting the data into vectors);
receiving recommendation feedback comprising lexicographic recommendation selections in response to the [selectable real-time lexicographic recommendations] (Lu in [0081] teaches the user may provide feedback on one or more of the predictions received).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for using an API. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Broidy, as modified above, does not teach, however Gaudiano teaches
selectable real-time lexicographic recommendations (Gaudiano in [0020] teaches a user interacting with a broader number of keywords presented in such interactive text clouds is more likely to find and select keywords that are of particular interest).
Gaudiano is considered to be analogous to the claimed invention because it is in the same field of using interactive text clouds. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Gaudiano to allow for using interactive text clouds. Motivation to do so would allow for the user to interact with the words, e.g., by selecting some words or topics (for instance by clicking on them), removing some unwanted words or topics, focusing on specific words or topics, changing the number of words or topics displayed, or altering the way in which words or topics are displayed (Gaudiano [0012]).
Regarding claim 19, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, teaches the selectable real-time lexicographic recommendations, the inmate communications tokens, the inmate communication transcripts, and inmate communications language model.
Broidy further teaches
further comprising:
generating [the selectable real-time lexicographic recommendations] by searching a [vector] index of [the inmate communications tokens derived from the inmate communication transcripts] (Broidy in [0025] teaches calls are transcribed shortly after being completed between callers and callees (or multiple parties), and the transcription and other metadata regarding the calls are stored in a database in a searchable and indexed form).
Broidy, as modified above, does not teach, however Hendry teaches
searching a vector index (Hendry in [0014] teaches selecting, by the computing device, each ranked suggestion whose vector-space angle between the search query and the ranked suggestion is less than a vector-space angle between the search query and a document associated with a higher-ranked suggestion).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Hendry to allow for selecting a model. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Claims 18 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Broidy, in view of Hendry, in view of Lu, in view of Gaudiano, in view of Sar Shalom et al. (US Patent Pub. No. 20210233520 A1), hereinafter Sar Shalom.
Regarding claim 18, Broidy teaches a processor-implemented method (Broidy in [0031] teaches a method for processing audio data and indexing the audio data in a searchable database, and in [0066] teaches using processors for doing the processing),
comprising:
receiving inmate communications data (Broidy in [0056-0057] teaches receiving call data when monitoring conversations by inmates);
transcribing the inmate communications data to yield a plurality of inmate communication transcripts (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like);
at least one inmate communications search application (Broidy in [0025] teaches calls are transcribed shortly after being completed between callers and callees (or multiple parties), and the transcription and other metadata regarding the calls are stored in a database in a searchable and indexed form, and users can perform searches for callers, callees, keywords, and/or other information in calls across the system, and in [0024] teaches making the resulting data searchable and accessible via a user interface);
a plurality of inmate communications transcripts (Broidy in [0024] teaches obtaining audio and metadata information from voice calls, generating textual transcripts from those calls, and making the resulting data searchable and accessible via a user interface, by converting audio data from one or more sources (such as a telecommunications provider) into searchable usable text transcripts, in particular, the system is useful with respect to monitoring conversations by inmates in jails, prisons, correctional facilities, and the like);
providing [selectable] real-time lexicographic recommendations in response to at least one inmate communications search query (Broidy in [0045-0047] teaches using a system which can search inmate call data and provide various ways to view the lexicon content, such as views at various granularities (year, month, week, day, hour, day of week, hour of day) to identify any patterns of behavior of the individuals or equipment),
the [selectable] real-time lexicographic recommendations comprising at least one of a synonym, slang term, mistranscription correction, and misspelling correction (Broidy in [0047] teaches using a system which can manage lexicon content (words, synonyms, aggregation hierarchy)).
Broidy teaches the plurality of inmate communications and the plurality of inmate communication transcripts.
Broidy does not teach, however Hendry teaches
tokenizing the [plurality of inmate communication transcripts] to yield a [plurality of inmate communications] tokens (Hendry in [0034] teaches parsing documents into tokens);
removing stop words from the [plurality of inmate communications] tokens (Hendry in [0034] teaches a “stop list” may also be utilized when parsing a document, where a stop list may direct the parser to ignore certain tokens that would not be useful to forming relevant search queries from the selected document, and grammatical articles such as “a”, “an”, and “the” are examples of tokens frequently in stop lists);
selecting a language model architecture for an inmate communications language model (Hendry in [0047-0050] teaches that many language models are available, and that particular language models suit a particular purpose better or worse than other language models depending on circumstances relating to the particular purpose, and a language model in the context of the present application determines how relevant a search query is for a particular document based on a frequency with which each term is present in the document and the frequency with which that term is present in other documents within the corpus, and applying a language model, the M most relevant search queries for a document can be identified).
Hendry is considered to be analogous to the claimed invention because it is in the same field of using language models to analyze text. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy further in view of Hendry to allow for parsing documents into tokens. Motivation to do so would allow for generating keys with common roots for search queries reflecting similar concepts, concept ordering allows queries reflecting similar concepts to be grouped together in a manner that allows for storage efficiency and for efficient implementation of spelling correction and search suggestions (Hendry [0051]).
Broidy, as modified above, teaches the plurality of inmate communications, at least one inmate communications search application, real-time lexicographic recommendations, tokens, inmate communications language model.
Broidy, as modified above, does not teach, however Lu teaches
performing word embeddings to map the [plurality of inmate communications tokens] to a corresponding [plurality of inmate communications] vectors (Lu in [0025] teaches generating embeddings corresponding to vector representations from the data);
splitting the [plurality of inmate communications] vectors into training vectors and testing vectors (Lu in [0048] teaches validation and testing are performed for a machine learning model, such as based on validation data and test data, as is known in the art, and [0025] teaches converting the [two types of] data into vectors);
training the [inmate communications language model] with the training vectors (Lu in [0048] teaches training data is used to train the machine learning model, and [0025] teaches converting the data into vectors);
validating the [inmate communications language model] with the testing vectors (Lu in [0048] teaches validation data is used to validate the machine learning model, and [0025] teaches converting the data into vectors);
integrating the [inmate communications language model] with [at least one inmate communications search application] via a model application programming interface (Lu in [0052] teaches using an API to integrate a server with a device);
receiving recommendation feedback in response to the [selectable real-time lexicographic recommendations] (Lu in [0081] teaches the user may provide feedback on one or more of the predictions received);
and
updating at least one [inmate communications language model] parameter based on the recommendation feedback (Lu in [0081] teaches the labeled training data for the machine learning model may be updated based on the feedback and the updated labeled training data may be used to re-train the machine learning model).
Lu is considered to be analogous to the claimed invention because it is in the same field of training a machine learning model. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Lu to allow for re-training a machine learning model. Motivation to do so would allow for a machine learning model to be trained to recognize latent relationships among payors based on their industry description, and in this manner, the machine learning model may better predict transactions as business or personal transactions due, at least in part, to being trained with the industry name embeddings (Lu [0047]).
Broidy, as modified above, does not teach, however Gaudiano teaches
selectable real-time lexicographic recommendations (Gaudiano in [0020] teaches a user interacting with a broader number of keywords presented in such interactive text clouds is more likely to find and select keywords that are of particular interest).
Gaudiano is considered to be analogous to the claimed invention because it is in the same field of using interactive text clouds. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Gaudiano to allow for using interactive text clouds. Motivation to do so would allow for the user to interact with the words, e.g., by selecting some words or topics (for instance by clicking on them), removing some unwanted words or topics, focusing on specific words or topics, changing the number of words or topics displayed, or altering the way in which words or topics are displayed (Gaudiano [0012]).
Broidy, as modified above, does not teach, however Sar Shalom teaches
wherein the misspelling correction is generated based at least on a word embedding similarity score and a phonetic distance measure (Sar Shalom in 0059] teaches a encoding engine calculates vector distances as measures of contextual similarity between the customer utterance token “key boots” and the agent utterance tokens, and calcualtes phonetic distances as measures of phonetic similarity between the customer utterance token “key boots” and the agent utterance tokens, and the transcription improver combines the vector distances and the phonetic distances into aggregate distances, where in this case, the aggregate distance equally weights the vector distances and the phonetic distances).
Sar Shalom is considered to be analogous to the claimed invention because it is in the same field of using vector distances. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Sar Shalom to allow for combining the vector distances and the phonetic distances into aggregate distances. Motivation to do so would allow for improving a transcription including identifying, in the transcription, reliable channel tokens of an utterance of a reliable channel and unreliable channel tokens of an utterance of an unreliable channel, and generating, using a machine learning model, a vector embedding for an unreliable channel token and vector embeddings for the reliable channel tokens (Sar Shalom [0003]).
Regarding claim 20, Broidy, as modified above, teaches the method of claim 1.
Broidy, as modified above, does not teach, however Sar Shalom teaches
further comprising:
generating the misspelling correction based at least on: a word embedding similarity score, and a phonetic distance measure (Sar Shalom in 0059] teaches a encoding engine calculates vector distances as measures of contextual similarity between the customer utterance token “key boots” and the agent utterance tokens, and calcualtes phonetic distances as measures of phonetic similarity between the customer utterance token “key boots” and the agent utterance tokens, and the transcription improver combines the vector distances and the phonetic distances into aggregate distances, where in this case, the aggregate distance equally weights the vector distances and the phonetic distances).
Sar Shalom is considered to be analogous to the claimed invention because it is in the same field of using vector distances. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Broidy, as modified above, further in view of Sar Shalom to allow for combining the vector distances and the phonetic distances into aggregate distances. Motivation to do so would allow for improving a transcription including identifying, in the transcription, reliable channel tokens of an utterance of a reliable channel and unreliable channel tokens of an utterance of an unreliable channel, and generating, using a machine learning model, a vector embedding for an unreliable channel token and vector embeddings for the reliable channel tokens (Sar Shalom [0003]).
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.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PAUL J. MUELLER whose telephone number is (571)272-1875. The examiner can normally be reached M-F 9:00am-5:00pm (Eastern).
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, Daniel C. Washburn can be reached at 571-272-5551. 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.
PAUL MUELLER
Examiner
Art Unit 2657
/PAUL J. MUELLER/Examiner, Art Unit 2657
/DANIEL C WASHBURN/Supervisory Patent Examiner, Art Unit 2657