DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA and is in response to communications filed on 8/03/2026 in which claims 1, 3-21 are presented for examination.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 8/03/2026 has been entered.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, 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, 3-21 are rejected under 35 U.S.C. 103 as being unpatentable over Shtilkind et al. US 20240311223 A1 (hereinafter referred to as “Shtilkind”) in view of Mielke et al. US 20230135179 A1 (hereinafter referred to as “Mielke”) and further in view of Gatchalian et al. US 20240144198 A1 (hereinafter referred to as “Gatchalian”).
As per claim 1, Shtilkind teaches:
A computer-implemented method for finding historically similar incidents in a computer system, the method comprising:
receiving a plurality of historical data objects corresponding to a plurality of previous incidents, each of the plurality of historical data objects indicating an occurrence of a previous incident and including a historical resolution text description, wherein the previous incidents disrupt or cause loss of operation, services, or functions of the computer system (Shtilkind, [0005] – Obtaining a plurality of incident logs, each incident log including a respective sequence of events, wherein obtaining a plurality of incident logs is interpreted as receiving a plurality of historical data objects. [0003] – These logs may include investigatory steps, root cause determinations, and resolutions found in the classified events into respective cluster spaces. [0138] – Incidents may be “defects”, “problems”, “difficulties”, etc. experienced by users);
…
extracting noun phrases from each of the historical resolution text descriptions (Shtilkind, [0184] – Various techniques may be used to carry out this block, such as extractive or abstractive summarization. Extractive summarization involves selecting the most relevant sentences or phrases from an event and assembling them into a shorter version);
applying topic modeling to the extracted noun phrases to assign each of the plurality of historical data objects to a cluster of a plurality of clusters (Shtilkind, [0197] – This can involve identifying the most frequent or semantically representative features in the cluster (e.g., by way of techniques such as feature importance or frequency analysis to extract the most prevalent keywords or phrases, topic modeling, sentiment modeling). For instance, a cluster of symptoms all relating to a user not being able to access a WIFI network may be labeled with the text “Wi-Fi___33 not working.”);
…
receiving a current data object indicating an occurrence of a current incident associated with a configurable item, the current data object including an incident description (Shtilkind, [0002] – Each log may include text representing user-provided, agent-provided, and/or automatically-generated descriptions of the problem, the investigatory steps taken to determine the nature of the problem, the root cause of the problem, and how the problem was resolved);
applying topic modeling to the extracted noun phrase of the current data object to assign the current data object to a particular cluster of the plurality of clusters (Shtilkind, [0003] – These models may include a classifier that categories events in the logs into one or more pre-defined classes, as well as a clustering model that groups similar symptoms, investigatory steps, root cause determinations, and resolutions found in the classified events into respective cluster spaces); and
Although Shtilkind teaches historical data, and resolutions as well as cluster space with vectors, Shtilkind doesn’t explicitly teach embeddings which may be different from clusters in a vector space, however, Mielke clearly teaches this gap:
generating a historical embedding of each of the plurality of historical data objects (Mielke, [0099] – The entity resolution module 212 may additionally extract features from contextual information, which is accessed from dialog history between a user and the assistant system 140. The entity resolution module 212 may further conduct global word embedding, domain-specific embedding, and/or dynamic embedding based on the contextual information);
storing the historical embedding and cluster for each of the plurality of historical data objects in an index (Mielke, [0095] – The AUM 354 may then index these data for retrieval as needed. Index ID may be created for such purpose. In particular embodiments, given an “index key” (e.g., PHOTO_LOCATION) and “index value” (e.g., “San Francisco”), the AUM 354 may get a list of memory IDs that have that attribute (e.g., photos in San Francisco));
generating a current embedding of the current data object (Mielke, [0099] – The entity resolution module 212 may further conduct global word embedding, domain-specific embedding, and/or dynamic embedding based on the contextual information. The entity resolution module 212 may communicate with different graphs 352 including one or more of the social graph, the knowledge graph, or the concept graph to extract ontology data that is relevant to the retrieved information from the context engine 220);
…
…
identifying a set of historically similar incidents by applying a Euclidean distance formula to the current embedding and at least a subset of the historical embeddings associated with the particular cluster (Mielke, [0331] – A similarity metric custom-character and custom-character may be a Euclidean distance); and
automatically implementing a corrective action to correct the current incident based on at least one of the historical resolution text descriptions of the set of historically similar incidents, thereby improving operation of the computer system (Shtilkind, [0138] – These incident logs are typically created by users who are experiencing technical difficulties, and are then assigned to virtual agents (e.g., chatbots) and/or human agents for resolution, wherein agents are automated which is interpreted as an automatic implementation. [0145] – Once the root cause is identified, a resolution is usually apparent. For example, if the root cause is that the user is attempting to connect to a Wifi access point that is too far away (and thus has low signal strength in the location of the user), the resolution might be for the user to move closer to the access point or connect to a closer access point. If the root cause is that the access point is low on memory (perhaps due to a memory leak defect in its firmware), the resolution might be for the agent to reboot the access point).
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify Shtilkind’s invention in view of Mielke in order to include explicit embedding of text; this is advantageous because it allows the text to be placed in an embedding space for precise classification purposes using accurate mathematical formulas like the Euclidean distance formula (Mielke, paragraph [0331]).
Although Shtilkind teaches selecting the most relevant sentences or phrases, Shtilkind doesn’t explicitly teach extracting noun phrases with respect to incident resolution, however, Gatchalian teaches:
extracting, using a natural language processing module, noun phrases from the current embedding, wherein the extracted noun phrases correspond to specific terms and concepts occurring in incident resolution (Keyword extraction may follow closing of a ticket associated with the service request… keyword extraction may be based on/follow detecting trigger phrases, within the communications, indicating that the service request has been resolved. For example, the trigger phrases may be “it works,” “no error,” “error has cleared,” and/or the like. The NLP module 162 may determine/analyze noun phrases and/or n-grams within the communications to determine presence of trigger phrases);
It would have been obvious for one of ordinary skill in the art prior to the effective filing date of the claimed invention to modify Shtilkind’s invention in view of Gatchalia in order to extract noun phrases that correspond to incident resolution; this is advantageous because it may enable determination of words and/or phrases that may correspond to symptoms associated with a service request, computing resources associated with the service request, error code associated with the service request, prospective solutions attempted for resolving the service request, whether the attempted prospective solution was successful or not for resolving the service request, etc. (Gatchalia, paragraph [0050]).
As per claim 3, Shtilkind as modified teaches:
The method of claim 1, wherein generating a historical embedding of each of the plurality of historical data objects includes:
processing the historical embedding (Mielke, [0044] – Any of the processing, detection, or keyword verification may be performed by the rendering device 137 and/or the companion device 138. [0092] – Parsing is interpreted as processing); and
applying a large language machine learning module to determine the historical embedding, wherein the historical embedding is a numerical representation of a corresponding historical resolution text description (Mielke, [0261] – The assistant system 140 may use a large language model (e.g., like GPT-3) as a chat bot/user simulator to perform QA test on assistant updates).
As per claim 4, Shtilkind as modified teaches:
The method of claim 1, wherein generating a current embedding of the current data object further includes:
processing the current embedding (Mielke, [0044] – Any of the processing, detection, or keyword verification may be performed by the rendering device 137 and/or the companion device 138. [0092] – Parsing is interpreted as processing. [0101] – The dialog intent resolution 356 may resolve the user intent associated with the current dialog session based on dialog history between the user and the assistant system); and
applying a large language machine learning module to determine the current embedding, wherein the current embedding is a numerical representation of the incident description (Mielke, [0261] – The assistant system 140 may use a large language model (e.g., like GPT-3) as a chat bot/user simulator to perform QA test on assistant updates).
As per claim 5, Shtilkind as modified teaches:
The method of claim 1, further including:
processing the current data object prior to applying the topic modeling, the processing including applying one or more of a lower casing algorithm, a tokenization algorithm, a punctuation mark removal algorithm, a stop word removal algorithm, a stemming algorithm, or a lemmatization algorithm (Shtilkind, [0191] – Hierarchical modeling (building a tree-like structure of clusters, where events are grouped together based on the similarity of their vectors), topic modeling (identifying the underlying topics or themes in the events based on word distribution).
As per claim 6, Shtilkind as modified teaches:
The method of claim 1, wherein the incident description of the current data object includes a text summary of a problem or issue associated with the configurable item (Shtilkind, [0173] – These techniques result in a summarization and characterization of past incident logs that can be used to guide how agents proceed while addressing similar incidents encountered in the future), and
the historical resolution text description is a record of steps taken to diagnose and fix a corresponding previous incidents (Shtilkind, [0002]-[0005] – Investigatory steps taken to determine the nature of the problem and root cause).
As per claim 7, Shtilkind as modified teaches:
The method of claim 1, wherein identifying a set of historically similar incidents includes:
determining a similarity score for each of the set of historically similar incidents based on the application of the Euclidean distance formula (Mielke, [0331] – A similarity metric custom-character and custom-character may be a Euclidean distance).
Claims 8-14 are directed to a system performing steps recited in claims 1-7 with substantially the same limitations. Therefore, the rejections made to claims 1-7 are applied to claims 8-14.
Claims 15-20 are directed to a non-transitory computer readable medium performing steps recited in claims 1-7 with substantially the same limitations. Therefore, the rejections made to claims 1-7 are applied to claims 15-20.
As per claim 21, Shtilkind as modified teaches:
The method of claim 1, further including:
preprocessing the historical resolution text descriptions, prior to the extracting the noun phrases (Shtilkind, [0002] – Preprocessing a corpus of logs to identify the problems, investigatory steps, root cause determinations, and ultimate problem resolutions therein),
by performing at least one lower casing, tokenization, punctuation mark removal, stop word removal, stemming, or lemmatization of the historical resolution text descriptions (Shtilkind, [0182] – Other types of preprocessing may include removal of punctuation and/or stop words, stemming, lemmatization, and so on).
Response to Arguments
Applicant’s arguments with respect to claims have been considered but are generally 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.
Gatchalian is now incorporated into the rejection, and also teaches clustering operations on keywords extracted by the NLP model in [0054] which contextualizes the noun phrases mentioned in [0052] in a way that would be obvious to one of ordinary skill in the art to apply topic modeling to extracted phrases and noun phrases. It should be noted that Shtilkind explicitly teaches the phrase “topic modeling” in [0192] as cited in the rejection which matches the exact claim language, but lacks tying that idea to noun phrase extraction. Now, with Gatchalian, that gap is closed, and the prior art of record sufficiently maps to the claimed limitations.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Gupta et al. 1 July 2023, “From ChatGPT to ThreatGPT: Impact of Generative AI in Cybersecurity and Privacy”, https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10198233
Platais et al. US 20250138970 A1, US 20250138971 A1, US 20250131032 A1 teaches aggregating and mapping incident characteristics into a daily profile (Abstract). Fidelity Information Services.
Somech et al. US 20180218734 A1 teaches evaluating a similarity between a conversation between two or more users and a set of keywords characterizing at least one project associated with a user of the two or more users, where the conversation is captured by sensor data (Abstract).
Burton et al. US 12210839 B1 teaches tuples of, e.g. noun phrases believed to correspond to salient named entities in the dependency structure of the sentence, and belonging to well-known categories like organizations, products, landmarks, persons, works of art, cardinal or ordinal numbers, money amounts, nationalities or religious or political groups, or geographical or political entities. Named entity recognition suitable for computing these representations could be provided by tree search and keyword search on a dependency parse of the sentence but may be more effectively computed using a more flexible transformer (e.g. DistilBERT) and special span identification tokens in the nature of approaches to extractive question answering or a less semantically powerful and less resource-intensive convolutional neural network (CNN) implementing the tok2vec approach, which computes token embeddings with long-term dependency learnability crudely approximated by learned convolutional filters in column 23, lines 21-40.
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Matthew J. Ellis whose telephone number is (571)270-3443. The examiner can normally be reached on Monday-Friday 8AM-5PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kavita Stanley can be reached at (571) 272-8352. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
September 25, 2026
/MATTHEW J ELLIS/Primary Examiner, Art Unit 2153
