DETAILED ACTION
This communication is in response to the Amendments and Arguments filed on 03/16/2026.
Claims 1, 2, 4-14, and 16-23 are pending and have been examined.
All previous objections/rejections not mentioned in this Office Action have been withdrawn by the examiner.
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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 12/30/2025, 02/05/2026, 03/16/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Arguments
Applicant's arguments filed 03/16/2026 have been fully considered but they are either not persuasive or are moot.
Applicant asserts on pgs 13-14 that Hranj does not teach chunking based on a “predetermined fixed-size, as expressly claimed”. The Examiner respectfully disagrees with this assertion. First, the claims recite “splitting the textual data into fixed-sized textual data chunks”, and there is no recitation that the sizes are predetermined. Additionally, there is no indication in the claims as to what is meant by a “fixed size”, i.e. a fixed number of characters, words, sentences, and/or section of text. Using the BRI of fixed size, separating a document into multiple sentences, as taught by Hranj [0015], reads on splitting the text into a fixed size chunk, because each segment is a sentence. Should the Applicant wish to have the fixed size be interpreted as having a specific number of sub-parts, or that the size is, in fact, a predetermined value prior to any parsing of the document, such language should be recited in the claims.
Applicant further asserts on pgs 15-16, that Hranj teaches semantic embeddings for similarity comparisons rather than prompt generation and feeding the prompt into a language model to produce an output representing a summarization of the chunk. The Examiner respectfully disagrees with this assertion. Hranj does teach the identification of similar semantic embeddings (see [0031]), however, this identification of similarity is used to identify semantic embeddings that should be prioritized for summarization (see [0032-3],[0040]), including which semantic embedding summaries should be presented first because they are the most important. Hranj specifically teaches in [0033-34] that a summarization engine (which can be a language model such as GPT), receives summarization instructions defining the summarization scope for each embedding, and presentation order of the summaries, along with the semantic embeddings that are to be summarized. The summarization engine then provides the requested summaries for the semantic embeddings in accordance with the summarization instructions. Therefore, Hranj teaches generating prompts for each textual data chunk and feeding the generated prompt into a specific-trained language model to output a comprehensive summarization of the textual data.
Regarding the arguments related to progressive filtering, Applicant’s arguments have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. Please see the updated mappings below citing Lukyanenko and Penfield for further detail.
Hence, Applicant’s arguments are either not persuasive or are moot.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1, 2, 4-14, and 16-23 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.
Claims 1, 12, 13, respectively, recite “the data chunk” in the second to last limitation. There is insufficient antecedent basis for this limitation in the claim.
Claims 2, 4-11, 14, and 16-23 are rejected as being dependent upon a rejected base claim.
The Examiner notes that dependent claims 7 and 19 also recite “the data chunk”, and should be amended to match the amendments made to claims 1 and 13, to ensure consistency in the claim terms.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1, 2, 4, 5, 7-14, 16, 17, and 19-23 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hranj et al. (US PG Pub No. 2024/0289366), hereinafter Hranj, in view of Lukyanenko et al. (U.S. PG Pub No. 2023/0063713), hereinafter Lukyanenko, and further in view of Penfield et al. (U.S. Patent No. 11,893,048), hereinafter Penfield.
Regarding claims 1, 12, and 13, Hranj teaches
(claim 1) A method for efficiently generating a call summary (method for providing summaries [0003]), the method comprising:
(claim 12) A non-transitory computer-readable medium having stored thereon instructions for causing a processing circuitry to execute a process (computer readable media storing instructions for access by a computing device to perform operations [0049],[0053],[0055]), the process comprising:
(claim 13) A system for efficiently generating a call summary, comprising:
(claim 13) a processing circuitry; and
(claim 13) a memory, the memory containing instructions that, when executed by the processing circuitry, configure the system to (a device including processing unit and memory storing instructions for access by the computing device to perform operations [0049-50],[0053],[0055]):
ingesting textual data and respective input data including topic data (a request for a summary of a document is received, and the document is retrieved, i.e. ingesting textual data, and additional input data is collected including contextual information and knowledge information about the user, i.e. ingesting…respective input data, where the context includes topics in the document, i.e. respective input data including topic data [0013-7],[0029]);
splitting the textual data into fixed-sized textual data chunks (the document is separated into multiple segments, i.e. splitting the textual data, such as a section, paragraph, or sentence, i.e. into fixed-sized textual data chunks [0015],[0029]);
formatting the textual data chunks and respective input data, …, into a unified data format, wherein the textual data chunks in the unified data format are in a same data format (the document is separated into multiple segments, i.e. textual data chunks, and a semantic embedding is generated for each segment, i.e. formatting the textual data chunks…into a unified data format wherein the textual data chunks in the unified data format are in a same data format, and where the level of similarity between semantic embeddings indicate whether different embeddings have a similarity in topic, i.e. formatting…respective input data into a unified data format [0015],[0017],[0029],[0031]);
generating a prompt for each textual data chunk …, wherein the prompt is generated from the formatted textual data chunks and the respective input data in the unified data format (the semantic embeddings, which include semantic embeddings for the context and knowledge information, are provided as input to a personal knowledge system for the user, i.e. the formatted textual data chunks and the respective input data in the unified data format, and the personal knowledge system outputs indications of the semantic embeddings that are to be summarized for the user ranked in order and indicating a summarization scope for each semantic embedding, i.e. generating a prompt for each textual data chunk of the textual data wherein the prompt is generated from the formatted textual data chunks…in the unified data format, and the embeddings may be prioritized according to the topic indicated by the semantic embedding, i.e. the prompt is generated from…the respective input data in the unified data format [0014-5],[0017],[0029],[0031-3]);
feeding each generated prompt to a …-trained language model to output a summary for the each textual data chunk, wherein the summary is a comprehensive summarization that describes the textual data of the data chunk (the summarization engine may be a model such as a GPT or other machine learning model, i.e. trained language model, that receives indications of the semantic embeddings to be summarized and corresponding summarization instructions, i.e. feeding each generated prompt to a trained language model, where the summarization engine generates summaries for each semantic embeddings that should be summarized according to the summarization scope, i.e. output a summary for each textual data chunk, where the summary may include a summary of each paragraph in the document, i.e. summary is a comprehensive summarization that describes the textual data of the data chunk [0012],[0032-4]); and
causing a display of the summary via a user device (the summaries are provided to the user device for display [0034]).
While Hranj provides prioritizing segments of text for summarization, Hranj does not specifically teach different stages of filtering textual data for further processing, and thus does not teach
progressively filtering the textual data chunks in at least a first filtering stage and a second filtering stage, …, wherein the second filtering stage is applied only to a subset of the textual data that pass the first filtering stage, and wherein the progressive filtering filters out textual data chunks associated with a predefined topic data;
…remaining after progressive filtering…;
a specific-trained language model to output a summary.
Lukyanenko, however, teaches progressively filtering the textual data chunks in at least a first filtering stage and a second filtering stage, …, wherein the second filtering stage is applied only to a subset of the textual data that pass the first filtering stage, and wherein the progressive filtering filters out textual data chunks associated with a predefined topic data (the model pipeline uses a pre-processing and filtering process to process the dialogue that has been broken down into a collection of sentences or groupings of words, i.e. progressively filtering the textual data chunks in at least a first filtering stage and a second filtering stage, where the pre-processing includes general text cleansing to remove artifacts and unwanted characters or words, i.e. a first filtering stage, and the series of filtering includes filtering out information noise filtering messages for greetings and/or approvals, filtering salutations, and/or other possible informational noise, i.e. wherein the progressive filtering filters out textual data chunks associated with a predefined topic data, and where each step of pre-processing and filtering uses the results of the previous filtering, i.e. the second filtering stage is applied only to a subset of the textual data that pass the first filtering stage [0013-5],[0017-9]);
…remaining after progressive filtering (a summary for a dialogue may be generated with the sentences after filtering, i.e. remaining after progressive filtering [0018-20])…;
a specific-trained language model to output a summary (models may be trained to generate summaries based on feedback from previous summaries, i.e. a specific-trained language model [0022]).
Hranj and Lukyanenko are analogous art because they are from a similar field of endeavor in using machine learning models to generate summaries. Thus, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the prioritizing segments of text for summarization teachings of Hranj with the filtering of different types of text and sentences before summarization as taught by Lukyanenko. It would have been obvious to combine the references to enable the use of only the most important sentences of a dialogue to generate the summary by filtering out messages that do not bring value (Lukyanenko [0014],[0018]).
While Hranj in view of Lukyanenko provides different stages of filtering, Hranj in view of Lukyanenko does not specifically teach that the filtering stages are performed using specific levels of resource-intensive processes, and thus does not teach
the first filtering stage is performed using a less resource- intensive filtering process and the second filtering stage is performed using a more resource-intensive filtering process.
Penfield, however, teaches the first filtering stage is performed using a less resource- intensive filtering process and the second filtering stage is performed using a more resource-intensive filtering process (text is pre-processed by an automated preprocessing module that performs tasks such as the removal of redundant information, i.e. first filtering stage is performed using a less resource- intensive filtering process, where the pre-processed text is then input into a pre-trained machine learning model designed to accept inputs of preprocessed text and output content that belongs to a specific field type, i.e. second filtering stage, and each specific machine learning model is fine-tuned on a custom dataset, i.e. the second filtering stage is performed using a more resource-intensive filtering process (6:66-7:20),(17:1-30),(18:29-19:1),(20:24-47)).
Hranj, Lukyanenko, and Penfield are analogous art because they are from a similar field of endeavor in processing text to produce specific outputs. Thus, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the different stages of filtering teachings of Hranj, as modified by Lukyanenko, with the use of a set of fine-tuned models after the text has been pre-processed as taught by Penfield. It would have been obvious to combine the references to enable a machine learning model to more easily process text in order to extract information required, such as each machine learning identifying and extracting content belonging to a specific field for which it was designed (Penfield (17:1-30),(18:29-47)).
Regarding claims 2 and 14, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13, and Hranj further teaches
the textual data includes at least one of:
transcript data, message data, an email, a short message service (SMS), and a chat log (the document may be an email, instant messages, i.e. chat log, a text message, i.e. a short message service (SMS), phone call data converted to text, i.e. transcript data, or voicemail data converted to text, i.e. message data [0012],[0029]).
Regarding claims 4 and 16, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13, and Hranj further teaches
the topic data is related to topics derived from the textual data (the document can be text information, i.e. derived from the textual data, where the topics are detected in the document, i.e. the topic data is related to topics derived from the textual data [0012],[0014],[0031-2]).
Regarding claims 5 and 17, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13, and Hranj further teaches
aggregating a plurality of data chunks to provide a context to at least a portion of a simplified transcript (semantic embeddings are compared using similarity measurements to determine a similarity in topic between a semantic embedding and the knowledge information, i.e. provide a context to at least a portion of a simplified transcript, where semantic embeddings relating to the topic are identified, i.e. aggregating a plurality of data chunks [0014],[0027],[0029],[0031-3]).
Regarding claims 7 and 19, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13, and Hranj further teaches
the prompt includes at least one of:
a command, the textual data of the data chunk, and background details (the summarization engine receives indications of semantic embeddings to be summarized, i.e. the textual data of the data chunk, and corresponding summarization instructions such as summarization scope, summarization amount, presentation order, and output mode, i.e. a command…and background details [0034]).
Regarding claims 8 and 20, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13, and Hranj further teaches
generating a brief of the summaries of the textual data chunks by canonizing the summaries into a standardized representation (a document summary is generated based on the entire document, and may have different scopes, such as a summary for each semantic embedding identified for summarization, i.e. summaries of the textual data chunks, where the summaries of the semantic embeddings may be presented in a specific order or format defined in the summarization instructions, i.e. generating a brief of the summaries…by canonizing the summaries into a standardized representation [0012],[0032-4],[0040]), wherein the standardized representation of the brief is any one of:
a simplified transcript and a communication brief (the document may be phone call data converted into text and then summarized, such as having a one-sentence summary of each paragraph, i.e. a simplified transcript, or a summary of detected documents or an overall theme of the document, i.e. communication brief [0012],[0029],[0033-4],[0040]).
Regarding claims 9 and 21, Hranj in view of Lukyanenko and Penfield teaches claims 8 and 20, and Lukyanenko further teaches
feeding the generated brief of the summary to train the specific-trained language model (models may be trained to generate summaries, i.e. train the specific-trained language model, based on feedback from previous summaries, i.e. feeding the generated brief of the summary to train [0022]).
Where the motivation to combine is the same as previously presented.
Regarding claims 10 and 22, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13, and Lukyanenko further teaches
the specific-trained language model is a language model that is trained specific to a customer using specific customer data (models may be trained to generate summaries, i.e. a language model that is trained, based on feedback from previous summaries, such as a live agent viewing the summary and providing feedback on the summary associated with a dialogue with a particular user, i.e. a customer using specific customer data, and the summarization engine may recreate the summary for that dialog and use the information to train a new model, where the summaries are related to a continuing or new chat with a user, i.e. the specific-trained language model…is trained specific to a customer using specific customer data [0011],[0021-2]).
Where the motivation to combine is the same as previously presented.
Regarding claims 11 and 23, Hranj in view of Lukyanenko and Penfield teaches claims 10 and 22, and Lukyanenko further teaches
the call summary is a sales call, and wherein the trained language model is trained on sales data of the customer (the chat that is summarized can be for sales assistance, i.e. the call summary is a sales call, where dialogues related to the services provided, such as sales, are recorded and summarized, where models may be trained to generate summaries, i.e. trained language model, based on feedback from previous summaries, such as a live agent viewing the summary and providing feedback on the summary associated with a dialogue with a particular user, i.e. customer, and the summarization engine may recreate the summary for that dialog and use the information to train a new model, where the summaries are related to a continuing or new chat with a user, i.e. trained on sales data of the customer [0010-2],[0021-2]).
Where Hranj specifically teaches that the documents are phone call data [0012].
And where the motivation to combine is the same as previously presented.
Claim(s) 6 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hranj, in view of Lukyanenko, in view of Penfield, and further in view of Asi et al. (U.S. PG Pub No. 2023/0360640), hereinafter Asi.
Regarding claims 6 and 18, Hranj in view of Lukyanenko and Penfield teaches claims 1 and 13.
While Hranj in view of Lukyanenko and Penfield provides output mode or formatting for summaries, Hranj in view of Lukyanenko and Penfield does not specifically teach that the summary is formatted into bullet points, and thus does not teach
the summary is formatted into a bullet point format.
Asi, however, teaches the summary is formatted into a bullet point format (the summary is in bullet points Fig. 2,[0021-3]).
Hranj and Asi are analogous art because they are from a similar field of endeavor in providing automated summaries to a user. Thus, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to modify the use of output modes and formatting for summaries teachings of Hranj with the use of bullet point lists for summaries as taught by Asi. It would have been obvious to combine the references to enable keyword-based groupings that remove redundant summarization points (Asi [0018]).
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 NICOLE A K SCHMIEDER whose telephone number is (571)270-1474. The examiner can normally be reached 8:00 - 5:00 M-F.
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, Pierre-Louis Desir can be reached at (571) 272-7799. 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.
/NICOLE A K SCHMIEDER/Primary Examiner, Art Unit 2659
Prosecution Insights