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 .
DETAILED ACTION
Claims 1-7, 9-19 are presented for examination.
This is Final Action.
Response to Arguments
Applicant's arguments filed 08/27/2026 have been fully considered but they are not persuasive.
Examiner has removed the objection in view of amended claims.
With respect to claims 1 and 19: applicant makes the following arguments:
1. The alleged Missing connection between category-matching failure and LLM invocation. Applicant explicitly argues, “First, the cited combination does not teach querying the LLM in response to determining that the string does not satisfy the matching criterion, and Li teaches away from doing so. Mondlock for its part, sends the user query to the LLM unconditionally within its pipeline (Mondlock, Figure 7C, Element 736) and never conditions the LLM call on a category-matching failure. The combination therefore fails to disclose the conditional querying recited in Claim 1.”
Examiner respectfully disagrees with the applicant. The rejection does not rely on Li as expressly disclosing an LLM call or Mondlock as independently disclosing the complete claimed conditional arrangement. Li, column 10, teaches determining that no category satisfies the classification threshold, and column 12 explains that obtaining a query-category assignment improves search results. Mondlock paragraph 40 teaches submitting a user query and predefined intents to an LLM and receiving identified intents as output. The proposed modification applies Mondlock’s additional classification technique to queries that remain unclassified after Li’s category evaluation, thereby obtaining category information unavailable from the initial classification stage. Queries already assigned a qualifying category can proceed without the additional interpretation. This is the asserted reason for connecting the failed classification determination to the LLM operation, rather than merely placing two unrelated operations in the same system.
2. Li’s conventional-search fallback allegedly excludes the proposed modification. Applicant explicitly argues “To the contrary, Li affirmatively directs that when no category score exceeds the threshold, the unmatched query is routed to a search engine without an assigned domain or query class.”
Examiner respectfully disagrees with the applicant. Li’s Column 14, example 2, does describe processing an unclassified query without a domain or query-class assignment. However, that passage does not state that additional classification is prohibited or must be avoided. Moreover, the proposed modification need not remove the conventional search operation: Li’s initial classification remains, Mondlock supplies additional interpretation when no qualifying category is obtained, and the POI search operation remains available to retrieve results. Applicant’s assertion that the modification necessarily eliminates Li’s search-engine functionality therefore does not describe this proposed arrangement. Li also expressly contemplates different configurations of initial classification and subsequent meta-classification, including threshold-controlled use of subsequent classification.
3. Claims 1 and 19: Applicant argues Mondlock allegedly supplies only retrieved documents and context. Applicant explicitly argues “Second, none of the cited references teaches providing the predetermined plurality of POI categories to the LLM. Mondlock is a retrieval-augmented generation pipeline that supplies the LLM with retrieved documents and context to ‘generate an answer’; the phase ‘POI category’ does not appear in Mondlock, and Mondlock never provides a predetermined POI-category set to the LLM.”
Examiner respectfully disagrees with the applicant. Applicant’s description does not account for Mondlock’s separate intent-classification disclosure. Paragraph 40 expressly states that the intent classification module may “submit the user query and the pre-defined intents to the LLM service 170” and “receive intents as output.” Mondlock therefore teaches supplying predefined classification alternatives together with the user query and obtaining classification-related output, rather than only supplying retrieved documents for answer generation. The rejection relies on adapting those predefined alternatives to POI categories in the combined POI-search system; it does not contend that Mondlock independently identifies its alternatives as POI categories.
4. Applicant argues, “Chen2 describes a training scheme for a prompt generator in which sampled class labels, namely, named-entity-types, are prepended to a frozen language model during training. Chen2 does not provide a predetermined plurality of POI categories to a LLM at inference time so as to direct the LLM’s output to a POI category label, as Claim 1 required.”
Examiner respectfully disagrees with the applicant. The characterization of Chen2’s label prompting as confined to training is not supported by its inference disclosure. Paragraph 70 teaches prompts containing label names and corresponding soft tokens; paragraph 71 teaches the input text prepended with the prompt to the pretrained language model; and paragraph 75 expressly describes providing the resulting task output during inference. Figure 7A likewise distinguishes the training and inference paths. Thus, Chen2 teaches label-containing prompts supplied to the model during inference. The additional application of those labels as POI-category alternatives remains the proposed modification to the POI-search system, not a training-only limitation of Chen2.
5. Applicant argues, “The office Action’s reliance on Chen2’s generic label-prompting to supply this limitation reads the specific POI-category context out of the claim. Chen2 is directed to generic text classification of provides no technical nexus to geospatial POI routing or map databases.”
Examiner respectfully disagrees with the applicant. The proposed combination retains the POI-category context. Chen in paragraph 147 expressly teaches category words, including “Hotel” and “Supermarket,” from individual names, including “Sheraton” and “Carrefour.” Li supplies classification against predefined categories, Mondlock supplies LLM classification using predefined alternatives, and Chen2 supplies label-containing prompts during inference. The proposed adaptation uses the POI-category alternatives of the combined search system as the alternatives supplied to the language model, so that classification output can guide retrieval of POIs of the relevant type. The technical connection is therefore the use of category classification to interpret a POI query and support POI retrieval – not assertion that generic class labels inherently are POI categories.
6. Applicant argues “Furthermore, the Examiner’s proposed modification relies on impermissible bodily incorporation, extracting the isolated terms ‘class labels’ and ‘language model’ from Chen2 merely because they match claim terminology.”
Examiner respectfully disagrees with the applicant. The relied-upon teaching in Chen2’s operational sequence of identifying the available labels, generating a label-containing prompt, supplying the prompt with the input text, and obtaining classification output during inference. The proposed POI application preserves those functional relationships: the input is the user’s query, the classification alternatives are POI categories, and the output identifies a category relevant to retrieval. Mondlock paragraph 40 independently supplies an LLM implementation that receives the query and predefined alternatives and returns classification output. The combination therefore relies on an identified classification technique rather than isolated matching terminology. It does not require transferring Chen’s learned soft-token values unchanged into Mondlock’s model; any model-specific implementation must remain consistent with the disclosed interface.
Applicants argument in view of claim 17 was persuasive hence claim 17 has been indicated as allowable.
Claims 19 is similar to claim 1 hence rebutted similarly.
Claims 2-7 and 9-16 are dependent on claim 1 hence similarly rejected.
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 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 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 1-7, 9-16 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Chen et al. (US2018/0349380 – IDS) in view of Mondlock et al. (US 2025/0131247) and of Li et al. (US 8,843,470) further in view of Chen et al (US 2023/0419049 – “Chen2”)
1. Chen teaches, A method for identifying at least one geospatial location based on user input (Paragraph 1 - teaches a point of interest may be any location to which a user may wish to navigate. Examples of points of interest includes… restaurants, hotels, retail stores, airports…, Chen), the method comprising:
using the string for identifying at least one point-of-interest (POI), wherein the string comprises (Paragraph 4 - teaches match a first text segment to a first point-of-interest segment index… match a second text segment… and use the … indices to identify one or more candidate POI entries, Chen):
identifying, as the at least one geospatial location, a geospatial location corresponding to the at least one POI (Abstract – teaches identify one or more candidate POI entities matching both the first and second text segments – thus disclosing identifying candidate POIs corresponding to user queries, which inherently correspond to physical locations, Chen);
Chen does not explicitly teach,
receiving, as the user input, a string from a user;
determining whether at least part of the string satisfies a matching criterion with at least one a predetermined plurality of POI categories;
in response to determining that the string does not satisfy the matching criterion, querying, based on at least part of the string and a predetermined plurality of POI categories, a large language model (LLM) for output relating to identification of at least one POI category label; and
…wherein the querying comprises providing the predetermined plurality of POI categories to the LLM.
However, Mondlock teaches,
receiving, as the user input, a string from a user (Fig 7A: 710 – receive a user query, Mondlock).
querying, based on at least part of the string and a predetermined plurality of POI categories, a large language model (LLM) for output relating to identification of at least one POI category label (Fig 7C: 736- teaches send augmented user query to LLM to generate an answer – thus disclosing sending a user query to LLM to obtain output generated from the query input, Mondlock)
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which said subject matter pertains to incorporate the LLM-based query processing of Mondlock in the query classification framework of Li within the POI search system of Chen in order to improve interpretation of user queries that do not satisfy predefined category matching criteria and thereby improve the identification of relevant POI and associated geospatial locations. The combination merely applies known natural-language interpretation techniques using a LLM to a conventional POI query processing system, yielding predictable improvement in interpreting ambiguous or unmatched user queries.
However, Li teaches,
determining whether at least part of the string satisfies a matching criterion with at least one a predetermined plurality of POI categories (Col 18: lines 8-20 - teaches generate category scores for meta-classifier categories… at least one category score being greater than a threshold value – thus disclosing determining whether a query satisfies a classification criterion by comparing category scores against a threshold, corresponding to determining whether the query matches one of a plurality of predefined categories, Li);
in response to determining that the string does not satisfy the matching criterion (Col 10: lines 14-24- teaches if none of the category scores is above a threshold value, then the query is not associated with any of the meta-classifier categories – thus disclosing determining when a query does not satisfy the classification criterion (i.e. no category score exceeds the threshold, Li).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which said subject matter pertains to modify Chen’s POI identification method to incorporate the query classification techniques of Li to determine whether a query matches predefined categories and, where the query does not satisfy the matching criteria, to utilize the LLM-based query processing of Mondlock to interpret the query and determine an appropriate category or response. Such modification represents the predictable use of known query-processing techniques to improve the interpretation of user queries and thereby improve retrieval of relevant POIs and associated geospatial locations.
However, Chen2 teaches,
…wherein the querying comprises providing the predetermined plurality of POI categories to the LLM (Abstract, Claim 1 – teaches a set of labels of interest is generated by sampling from a set of possible labels…. Prompting data that comprises a first plurality of class labels is generated and a pretrained language model generates a text output is response to each of the generated prompts – thus disclosing providing a predetermined labels/categories to a language model as part of the model prompt/input so that the model output related to identification of a label/class, Chen2. In modified POI system of Chen, the predetermined labels/classes correspond to the predetermined plurality of POI categories provided to Mondlock’s LLM).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which said subject matter pertains to modify the combination of Chen/Li/Mondlock with Chen2 because Chen2 teaches providing a plurality of possible class labels to a language model as part of prompt/input for text classification. Since Chen/Li/Mondlock combination uses an LLM to interpret a user query when category matching fails, a POSITA would have been motivated to provide Chen’s predetermined POI categories to the LLM using Chen2’s label-prompting technique so that the LLM output is directed to identifying one of the predetermined POI category labels. This would improve classification accuracy and constrain the LLM output to the known POI category taxonomy, yielding predictable results.
2. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, wherein using the string further comprises:
extracting, from the output of the LLM, at least one candidate POI category label and identifying the at least one POI based on the at least one candidate POI category label (Fig 7C:736 - teaches send augmented user query to LLM to generate an answer, Mondlock).
3. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 2, wherein identifying the at least one POI based on the at least one candidate POI category label comprises:
in response to determining that the at least one candidate POI category label corresponds to at least one corresponding POI category from among a predetermined plurality of POI categories (Paragraphs 53 and 57 - teaches comparing candidate labels against a fixed taxonomy of POI categories, Chen), identifying the at least one POI based on the at least one corresponding POI category (Paragraphs 53-59 - teaches identifying POIs based on the matched categories, Chen).
4. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 3, further comprising, in response to determining that the at least one candidate POI category label does not correspond to at least one POI category from among the predetermined plurality of POI categories (Paragraph 52 – teaches where a candidate string/label has no correspondence in the known POI categories/entries, Chen), identifying the at least one POI by searching a POI database and/or map database based on the candidate POI category label (Paragraph 143 – teaches …no corresponding index may be found… which may indicate that the segment does not appear in any known point-of-interest entry (process then continues with retrieval techniques) - describes a fallback handling when no matching index/category exists, leading the system to search for candidate POIs in the database, Chen).
5. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 4, further comprising providing data for displaying the at least one POI on a display (Paragraph 1 - teaches navigation system allow a user to search for POI and present the results to the user though navigation interface – thus disclosing returning POI search results and presenting them to the user though the navigation interface, which necessarily involves providing data for displaying POIs on a display device, Chen), wherein each POI displayed on the display is selectable by the user (Paragraph 2 – the navigation system may prompt the user to confirm that the identified POI is indeed what the user intended, Chen).
6. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 5, further comprising, in response to receiving a selection of a POI from among the POI displayed on the display (Paragraph 2 – the navigation system may prompt the user to confirm that the identified POI is indeed what the user intended, Chen) that was identified by searching a POI database and/or map database (Paragraph 4 - teaches match a first text segment to a first POI segment index stored in the database… and use the indices to identify one or more candidate POI interest entries, Chen )based on the candidate POI category label (Col 18: lines8-20 - teaches generate category scores for meta-classifier categories… select meta-classifier category based on the category score greater than the threshold value, Li), adding, to the POI database, data indicative of an association between the selected POI and the corresponding candidate POI category label (Paragraph 95, Fig 3:340 - teaches the system maintains POI data structures and updates or stores indexed POI entries associated with query processing).
7. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 3, further comprising, in response to determining that the at least one candidate POI category label does not correspond to at least one POI category from among the predetermined plurality of POI categories (Paragraph 143 – teaches the system cannot find a correspondence between the input (candidate label) and existing POI categories, Chen), providing data for displaying, on a display, the at least one candidate POI category label, each POI category label displayed on the display being selectable by the user (Paragraphs 109 and 137 – teaches providing candidate output to the user for selection, Chen).
9. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 8, further comprising, in response to determining that at least part of the string satisfies the matching criterion for at least one POI category, identifying the at least one POI based on the at least one POI category (Paragraphs 53 and 58 – teaches matching part of user’s string to a stored category index and once categories/indices are matched, the system identifies POIs associated with them and returns the entries, Chen).
10. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 8, wherein determining whether at least part of the string satisfies a matching criterion with at least one of the predetermined plurality of POI categories comprises at least one of:
identifying, for each POI category label, whether there is a correspondence between the POI category label and at least part of the string; calculating, for each POI category among the predetermined plurality of POI categories, a similarity parameter between the string and the POI category (Paragraphs 53 & 109 – teaches testing between user input segment and POI category indices (labels) and calculating of similarity parameters between the input string and stored POI categories/entries, Chen); and
converting at least part of the received string into a search vector, calculating distances between the search vector and vector representations of each POI category among the predetermined plurality of POI categories (Paragraph 38 – teaches generating vector embeddings of textual input using techniques such as Word2Vec. Word2Vec represents words as vectors in a semantic vector space such that semantic similarity between textual input can be determined based on the distance between their corresponding vector representations. Thus, teaching converting a received string into a search vector and comparing the vector-to-vector representation of textual categories, Mondlock)
11. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, wherein using the string further comprises, in response to the output from the LLM indicating a failure of the identification of at least one candidate POI category label (Paragraph 42 – teaches recognition/identification failure analogous to an LLM being unable to identify a candidate label, Chen), identifying the at least one POI by searching a POI database and/or map database based on at least part of the string (Paragraph 143 – teaches fallback, if no label/index is matched, the system searches in the POI/map database using the input string segment, Chen).
12. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, further comprising:
providing data for displaying the at least one geospatial location to a user (Paragraph 1 - teaches a point of interest, may be any location to which a user may wish to navigate. Examples of points of interest includes… restaurants, hotels, retail stores, airports…, Chen), wherein:
providing data for displaying the at least one geospatial location to a user comprises providing data for visually indicating, for at least one geospatial location, that the geospatial location was identified based on a POI category (Paragraph 109 – teaches presenting candidate POIs grouped/ranked by category information, which serves as a visual indication to the user that the POI was identified via a category, Chen).
13. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, further comprising:
providing data for displaying the at least one geospatial location to a user, wherein: when a plurality of geospatial locations is identified (Paragraph 1 - teaches a point of interest may be any location to which a user may wish to navigate. Examples of points of interest includes… restaurants, hotels, retail stores, airports…, Chen), providing data for displaying the plurality of geospatial locations to a user comprises providing data for grouping and/or ranking the plurality of geospatial locations that are identified in accordance with a corresponding POI category (Paragraph 109 – teaches grouping/ranking candidate POIs for presentation with the ranking tired to the recognition/category scoring, Chen ).
14. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, further comprising:
providing data for displaying the at least one geospatial location to a user, the providing including providing data for displaying, on a display, an indication of at least one POI category identified using the string (Paragraph 109 – teaches presenting recognized categories/entries to the user, which serves as indications of categories identified from the string, Chen), each indication of a POI category displayed on the display being selectable by the user (Paragraph 137 – teaches presenting candidate categories/entries in a user selectable format, Chen); and in response to receiving a selection of a given indication of a POI category, identifying at least one geospatial location based on a corresponding POI category and providing data for displaying the at least one geospatial location to the user (Paragraphs 58-59 – teaches once a user has selected, the system identifies POIs associated with the corresponding category and returns/displays then, Chen).
15. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, further comprising:
providing data for displaying the at least one geospatial location to a user, wherein: when a plurality of geospatial locations are identified, providing data for displaying the plurality of geospatial locations to a user (Paragraph 1 - teaches a point of interest may be any location to which a user may wish to navigate. Examples of points of interest includes… restaurants, hotels, retail stores, airports…, Chen) comprises providing data for displaying, on a display, an indication of at least one POI category corresponding to the at least one geospatial location (Paragraph 109 – teaches presenting candidate POIs grouped/ranked with category indications tied to their recognition source, Chen), each indication of a POI category displayed on the display being selectable by the user (Paragraph 137 – teaches presenting candidate POIs/categories in a user-selectable format, Chen).
16. The combination of Chen, Mondlock, Li and Chen2 teaches, The method of claim 1, further comprising:
providing data for displaying, on a display, the at least one geospatial location to the user, each geospatial location displayed on the display being selectable by the user (Paragraph 1 - teaches a point of interest may be any location to which a user may wish to navigate. Examples of points of interest includes… restaurants, hotels, retail stores, airports…, Chen); and
responsive to receiving the selection of at least one geospatial location, providing data indicative of instructions for navigating to the selected geospatial location (Paragraph 1 - teaches a point of interest may be any location to which a user may wish to navigate. Examples of points of interest includes… restaurants, hotels, retail stores, airports…, Chen).
Claim 19 is similar to the combination of claims 1 hence rejected similarly.
Allowable Subject Matter
Claims 17-18 is objected to as being dependent upon a rejected base claim but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Kallman et al. (US 11,809,508) – teaches artificial intelligence geospatial search (Abstract, Figs 2-6).
THIS ACTION IS MADE FINAL. 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 AMRESH SINGH whose telephone number is (571)270-3560. The examiner can normally be reached Monday-Friday 8am-5pm.
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, Ann J. Lo can be reached at (571) 272-9767. 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.
/AMRESH SINGH/Primary Examiner, Art Unit 2159
