DETAILED ACTION
This communication is in response to the Application filed on 02/25/2026. Claims 1-20 are pending and have been examined.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 13, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefore, subject to the conditions and requirements of this title.
Claims 1-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1, 11, and 19 recite A computer-implemented method for performing automated generation of descriptive metadata, the computer-implemented method comprising: receiving (i) one or more taxonomies associated with a media content ontology domain, (ii) a description of the one or more taxonomies, and (iii) one or more hierarchical relationships associated with the one or more taxonomies; generating an ontology based on the one or more descriptions and the one or more hierarchical relationships; generating a prompt that includes (i) a representation of the ontology, (ii) contextual information associated with a media content item, and (iii) a textual instruction to a [machine learning model]; and generating, via the machine learning model and based at least on the prompt, one or more descriptive metadata tags associated with the media content item.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind can perform each step of the method. A human can receive taxonomy information including descriptions and relationships on a piece of paper. For example, a drawn-out tree connecting information pertaining to a piece of media content. They can use the information to create their version of an ontology on a separate piece of paper. This could then be used to form a written question (prompt) to another human which would use the information to create labels for the media content. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims recite the additional component of a machine learning model. The machine learning model is merely being used to apply the mental process via a generic computing device. The machine learning model is described in paragraph 69 of the specification with a generic description of the component. Claim 11 specifically lists the additional components of non-transitory computer-readable media and processors. The CRM is merely being used to apply the mental process via a generic computing device. The CRM is described in paragraph 20 of the specification with a generic description of the component. The processor is merely being used to apply the mental process via a generic computing device. The processor is described in paragraph 17 of the specification with a generic description of the component. Claim 19 specifically lists the additional components of memories. The memory is merely being used to apply the mental process via a generic computing device. The memory is described in paragraph 21 of the specification with a generic description of the component. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 2 and 12 recite wherein generating the prompt comprises compressing, via one or more symbolic logic techniques, one or more of the ontology, the description of the one or more taxonomies, or the hierarchical relationships.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind can use symbol logic to compress information such as removing redundant words, replacing words with symbols, or adding words to show a connection between pieces of information. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 3 and 13 recite wherein the machine learning model includes a pre-trained generative machine learning model.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind can make the design decision to use a pre-trained MLM instead of a general purpose one. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 4, 14, and 20 recite wherein the prompt further includes example descriptive metadata tags associated with one or more example media content items and contextual information associated with the one or more example media content items.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of forming a prompt that includes metadata tags such as an actor’s name or a genre. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 5 and 15 recite wherein the media content ontology domain includes at least one of a film, a serialized television episode, a podcast, or a recording of a live audiovisual performance.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of understanding and creating labels for media content such as a film, a serialized television episode, a podcast, and a recording of a live audiovisual performance. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 6 and 16 recite wherein the media content ontology domain includes a film, and the one or more taxonomies include a genre, setting, or story element.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of understanding information such as a genre, setting, or story element. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 7 and 17 recite wherein the contextual information includes one or more of a script, a synopsis, a closed-caption file, a theatrical trailer, or a published review.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of understanding context information such as a script, synopsis, closed-captions, trailer, or published review. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claims 8 and 18 recite wherein each of the one or more descriptive metadata tags includes a value associated with one of the one or more taxonomies or a value associated with one of the one or more hierarchical relationships.
The limitations in these claims, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind can create labels that associate media content with information in a taxonomy and the hierarchical relationships. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claims are not patent eligible.
Claim 9 recites further comprising fine-tuning the machine learning model using training data including at least one of ontologies, source materials, or ground truth descriptive metadata tags associated with one or more training content items.
The limitations in this claim, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind can make the design decision of which type of data to train a model with. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim 10 recites wherein the machine learning model includes a multimodal machine learning model operable to accept two or more of still images, text, audio recordings, or video sequences as input.
The limitations in this claim, as drafted, are a process that, under broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The human mind is capable of understanding information found in images, text, audio recordings and video sequences. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
This judicial exception is not integrated into a practical application. The claims do not recite any additional components that were not present in the independent claims. Accordingly, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Claim Rejections - 35 USC § 102
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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-3, 5, 6, 8, 11-13, 15, 16, and 18-20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by “Improving Language Model Predictions via Prompts Enriched with Knowledge Graphs” (Brate et al.).
Regarding Claims 1, 11, and 19, Brate et al. teaches A computer-implemented method for performing automated generation of descriptive metadata, the computer-implemented method comprising:
(The basic idea of our method is to use the information about entities in KGs to expand cloze-style prompts with richer entity descriptions.) (Section 2, Paragraph 1).
Claim 11 states the alternative One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of:
(Table 2 lists the recall@n scores for each of the prompts described in Section3.2, for the BERT and RoBERTa large models respectively. For a given model and prompt, recall@1 and recall@5 values for each movie are calculated as the fraction of movie ground-truth genres predicted in the highest ranked n PLM mask predictions. The aggregated recall@n values reported in Table 2 are the micro-averaged recall@n scores across all movies in the test dataset, with respect to the model and prompt referenced.) (Section 4.2, Paragraph 1).
The models execute a test data set and the results are compiled in Table 2. This implies the use of such computer hardware within the system.
Claim 19 states the alternative A system comprising: one or more memories storing instructions; and one or more processors for executing the instructions to:
(Table 2 lists the recall@n scores for each of the prompts described in Section3.2, for the BERT and RoBERTa large models respectively. For a given model and prompt, recall@1 and recall@5 values for each movie are calculated as the fraction of movie ground-truth genres predicted in the highest ranked n PLM mask predictions. The aggregated recall@n values reported in Table 2 are the micro-averaged recall@n scores across all movies in the test dataset, with respect to the model and prompt referenced.) (Section 4.2, Paragraph 1).
The models execute a test data set and the results are compiled in Table 2. This implies the use of such computer hardware within the system.
receiving (i) one or more taxonomies associated with a media content ontology domain,
(The auxiliary data for each movie is extracted from Wikidata. This is done in a simplistic two-step-process using SPARQL queries.) (Section 3.1, Paragraph 1).
(First, the movies are linked to their respected Wikidata entities by IMDb or TMDB ID utilizing the Wikidata properties IMDb ID (wdt:P345) and TMDb movie ID (wdt:P4947). If this does not provide an entity, an exact string matching given the title is attempted as well.) (Section 3.1, Paragraph 2).
The method receives information related to a movie from Wikidata which includes movie websites such as IMDB and TMDb.
(ii) a description of the one or more taxonomies,
Fig. 1 shows the knowledge graph of information retrieved from Wikidata which represents a taxonomy.
and (iii) one or more hierarchical relationships associated with the one or more taxonomies;
Hierarchical relationships are shown in the form of the branches of the knowledge graph connecting to a single movie where each branch represents a labelled category.
generating an ontology based on the one or more descriptions and the one or more hierarchical relationships;
(The auxiliary data for each movie is extracted from Wikidata. This is done in a simplistic two-step-process using SPARQL queries. The queries operate on a batch of input records to reduce the number of requests and avoid timeout errors.) (Section 3.1, Paragraph 1).
The formation of an ontology is represented by the knowledge graph of the specific movie a query is being created rather than the larger database that is Wikidata.
generating a prompt that includes (i) a representation of the ontology,
(We enrich the naive prompt, for example Die Hard is of genre [MASK], through matching the movie Die Hard to the corresponding Wikidata item and extract auxiliary knowledge with SPARQL queries, and generating an enriched prompt using this auxiliary data.) (Section 3, Paragraph 1).
(The prompts are of the form: “<title> is a movie, of the genre [MASK].”, where is an aggregation of movie properties and corresponding values extracted from Wikidata pertaining the title in question, in some natural language format.) (Section 3.2, Paragraph 1).
Information from the ontology is included in the prompt. This can be seen in Fig. 1.
(ii) contextual information associated with a media content item,
(The prompts are of the form: “<title> is a movie, of the genre [MASK].”, where is an aggregation of movie properties and corresponding values extracted from Wikidata pertaining the title in question, in some natural language format.) (Section 3.2, Paragraph 1).
Context information related to the movie is included in the prompt.
and (iii) a textual instruction to a machine learning model;
(The prompts are of the form: “<title> is a movie, of the genre [MASK].”, where is an aggregation of movie properties and corresponding values extracted from Wikidata pertaining the title in question, in some natural language format.) (Section 3.2, Paragraph 1).
Fig. 1 shows the original text prompt to the model the enhanced version of it, both of which show the textual instruction of requesting the genre of the movie.
generating, via the machine learning model and based at least on the prompt, one or more descriptive metadata tags associated with the media content item.
(We then use both (a) the naive prompts and (b) the KG-enriched prompts to query various language models, and compare their performance on the entity genre classification task.) (Section 3, Paragraph 2).
Language models such as BERT or RoBERTa use the enhanced prompts to output a movie genre which represents a metadata tag for a media content item.
Regarding Claims 2 and 12, Brate et al. teaches the method of claims 1 and 11.
Brate et al. further teaches wherein generating the prompt comprises compressing, via one or more symbolic logic techniques, one or more of the ontology, the description of the one or more taxonomies, or the hierarchical relationships.
(Overall, a set of 28 properties was extracted and investigated for each entity. A simplified version of the utilized SPARQL query is given in 2. This query can easily be adapted to query other properties by adding these properties to the property values. From this set of properties, a subset of 10 manually selected domain-specific properties are used to constract the enriched prompts.) (Section 3.1, Paragraph 3).
The system extracts a subset from the Wikidata, thus compressing the overall amount of data. The logic used to compress this is shown in the SPARQL code shown in Section 3.1 where specific data property ID’s are pulled from the websites.
Regarding Claims 3 and 13, Brate et al. teaches the method of claims 1 and 11.
Brate et al. further teaches wherein the machine learning model includes a pre-trained generative machine learning model.
(In order to test our approach, we use the BERT [1] and RoBERTa large [20] pre-trained models.) (Section 4.1, Paragraph 1).
Pre-trained models such as BERT and RoBERTa are used to process the queries.
Regarding Claims 5 and 15, Brate et al. teaches the method of claims 1 and 11.
Brate et al. further teaches wherein the media content ontology domain includes at least one of a film, a serialized television episode, a podcast, or a recording of a live audiovisual performance.
(First, the movies are linked to their respected Wikidata entities by IMDb or TMDB ID utilizing the Wikidata properties IMDb ID (wdt:P345) and TMDb movie ID (wdt:P4947).) (Section 3.1, Paragraph 2).
The ontology domain is related to movies/films.
Regarding Claims 6 and 16, Brate et al. teaches the method of claims 1 and 11.
Brate et al. further teaches wherein the media content ontology domain includes a film, and the one or more taxonomies include a genre, setting, or story element.
(As a result, we obtain e.g. Die Hard is a movie, starring Bruce Willis, directed by John McTier- nan, of the genre [MASK].) (Section 3, Paragraph 1)
The ontology domain is specified as a movie. The specification that the media content is a movie can be considered a story element as opposed to a t.v. show it implies a longer time format.
Regarding Claims 8 and 18, Brate et al. teaches the method of claims 1 and 11.
Brate et al. further teaches wherein each of the one or more descriptive metadata tags includes a value associated with one of the one or more taxonomies or a value associated with one of the one or more hierarchical relationships.
(As a result, we obtain e.g. Die Hard is a movie, starring Bruce Willis, directed by John McTier- nan, of the genre [MASK].) (Section 3, Paragraph 1)
The genre (metadata tag) is associated with taxonomy values and hierarchical relationships as it is the genre of the movie from which the knowledge graph and properties are associated with.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 4, 7, 10, 14, 17, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over “Improving Language Model Predictions via Prompts Enriched with Knowledge Graphs” (Brate et al.) in view of “Transformative Movie Discovery: Large Language Models for Recommendation and Genre Prediction” (Raj et al.).
Regarding Claims 4, 14, and 20, Brate et al. teaches the method of claims 1, 11, and 19.
Brate et al. does not explicitly teach: wherein the prompt further includes example descriptive metadata tags associated with one or more example media content items and contextual information associated with the one or more example media content items.
However, Raj et al. teaches wherein the prompt further includes example descriptive metadata tags associated with one or more example media content items and contextual information associated with the one or more example media content items.
(Instead of requiring vast amounts of data for every specific task, the model is fine-tuned or prompted with only a few examples or instructions. This approach leverages the model’s pre existing knowledge, enabling rapid adaptation to new tasks or contexts.) (Section VI, Subsection 2, Paragraph 1).
Raj et al. teaches using few-shot examples within a prompt to provide a desired output. Figs. 7 and 8 show this being done to generate a summary. In this example the summary acts as the metadata tags and the prompt requests a summary from a movie transcript while providing an example transcript-summary pair from another movie.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the metadata tagging method as taught by Brate et al. to include examples within the prompt as taught by Raj et al. This would have been an obvious improvement to help guide the models’ results to what the user/system desires (Raj et al., Section VI, Subsection 2, Paragraph 1).
Regarding Claims 7 and 17, Brate et al. teaches the method of claims 1 and 11.
Brate et al. does not explicitly teach: wherein the contextual information includes one or more of a script, a synopsis, a closed-caption file, a theatrical trailer, or a published review.
However, Raj et al. teaches wherein the contextual information includes one or more of a script, a synopsis, a closed-caption file, a theatrical trailer, or a published review.
(Every movie has a movie ID, a multi-hot genre, whereas every user has a user ID, age, and occupation information associated with it. We have extracted the audio subtitles for each movie’s trailer via youtube-transcript-api and Youtube-Whisperer.) (Section IV, Subsection A).
(We plan to extend our evaluation to larger datasets like MovieLens-1M or Amazon Reviews, which contain a larger number of users and items.) (Section VIII, Paragraph 4).
Raj et al. shows various models receiving context information which include a script and synopsis (Figs. 7 and 8), closed-captions (subtitles), trailers, and reviews.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the metadata tagging method as taught by Brate et al. to include context information such as a script as taught by Raj et al. This would have been an obvious improvement to provide added context for the media that the metadata tags are made from (Raj et al., Section I, Paragraph 4).
Regarding Claim 10, Brate et al. teaches the method of claim 1.
Brate et al. does not explicitly teach: wherein the machine learning model includes a multimodal machine learning model operable to accept two or more of still images, text, audio recordings, or video sequences as input.
However, Raj et al. teaches wherein the machine learning model includes a multimodal machine learning model operable to accept two or more of still images, text, audio recordings, or video sequences as input.
(Our approach uses Large Language Models (LLMs) to generate movie summaries and predict genres based on audio subtitles from movie trailers, even for movies with limited user interaction data.) (Section III, Subsection B, Paragraph 1).
(Every movie has a movie ID, a multi-hot genre, whereas every user has a user ID, age, and occupation information associated with it. We have extracted the audio subtitles for each movie’s trailer via youtube-transcript-api and Youtube-Whisperer.) (Section IV, Subsection A).
The system of Raj et al. begins with the audio from movie trailers, which represents audio and/or video input. That audio is converted to a transcript, thus text is also a form of input.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the metadata tagging method as taught by Brate et al. to include inputs of various modalities as taught by Raj et al. This would have been an obvious improvement to enable inputs that match the media that the metadata is being (Raj et al., Section I, Paragraph 4).
Claims 9 are rejected under 35 U.S.C. 103 as being unpatentable over “Improving Language Model Predictions via Prompts Enriched with Knowledge Graphs” (Brate et al.) in view of US Patent Publication US 20230325428 A1 (Vartakavi et al.).
Regarding Claim 9, Brate et al. teaches the method of claim 1.
Brate et al. does not explicitly teach: further comprising fine-tuning the machine learning model using training data including at least one of ontologies, source materials, or ground truth descriptive metadata tags associated with one or more training content items.
However, Vartakavi et al. teaches further comprising fine-tuning the machine learning model using training data including at least one of ontologies, source materials, or ground truth descriptive metadata tags associated with one or more training content items.
(Without limitation, such a text classifier could be a machine-learning model trained with sentence structures labeled as being relevant or irrelevant. Once trained, the text classifier may receive as input the text representations of various podcast episodes and may evaluate each sentence in the text representations to determine a probability that the sentence is relevant, based on the earlier training, and the computing system may then establish a binary prediction of that probability.) (Paragraph 25).
(For instance, the computing system could do so using machine-learning-based role-identification similar to the NER process noted above. By way of example, a machine-learning model could be trained using a training dataset that labels named entities in text representations of podcast episodes as “Person - Host”, “Person - Guest”, or “Person - Subject”. Given the text representation of a podcast episode as input, possibly along with a prediction of a given text span being a celebrity name as discussed above, this machine-learning model may predict a role of that particular celebrity with respect to the podcast episode, such as whether the celebrity is a host of the podcast episode, a guest of the podcast episode, and/or a subject of the podcast episode, possibly outputting and storing binary indications as probabilities for each such role.) (Paragraph 41).
Vartakavi et al. teaches a method for finding metadata within a podcast that specifically states using ground-truth examples as training data for the model that generates the metadata.
It would have been obvious to a person having ordinary skill in the art before the time of the effective filing date of the claimed invention of the instant application to modify the metadata tagging method as taught by Brate et al. to include examples within the training data as taught by Raj et al. This would have been an obvious improvement to train the model on data that is relevant to the task it’s performing (Vartakavi et al., Paragraph 25).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICHOLAS DANIEL LOWEN whose telephone number is (571)272-5828. The examiner can normally be reached Mon-Fri 8:00am - 4:00pm.
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, Paras D Shah can be reached at (571) 270-1650. 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 D LOWEN/Examiner, Art Unit 2653
/Paras D Shah/Supervisory Patent Examiner, Art Unit 2653
08/28/2026
Prosecution Insights