US 12,169,497 B2
System and methods for context aware searching
Domingo Huh, Toronto (CA); Julian Brooke, Vancouver (CA); Elnaz Davoodi, Toronto (CA); and Jack G. Conrad, Eagan, MN (US)
Assigned to Thomson Reuters Enterprise Centre GmbH, Zug (CH)
Filed by Thomson Reuters Enterprise Centre GmbH, Zug (CH)
Filed on Nov. 19, 2021, as Appl. No. 17/531,693.
Application 17/531,693 is a continuation of application No. 16/181,729, filed on Nov. 6, 2018, granted, now 11,222,027.
Claims priority of provisional application 62/582,571, filed on Nov. 7, 2017.
Prior Publication US 2022/0083560 A1, Mar. 17, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 16/24 (2019.01); G06F 16/2457 (2019.01); G06F 16/38 (2019.01); G06F 16/93 (2019.01)
CPC G06F 16/24578 (2019.01) [G06F 16/38 (2019.01); G06F 16/93 (2019.01)] 16 Claims
OG exemplary drawing
 
1. A method of refining search results, the method comprising:
prior to receiving a user query:
automatically assigning, by a concept marker assigner of a server comprising a processor and a memory, one or more concept markers from a first list of master concept markers to each document of documents in a database comprising a collection of documents, wherein assigning the one or more concept markers comprises processing metadata in each document of the documents to identify concepts in the documents;
in response to receiving the user query from a user terminal:
ranking, by a concept marker recommender of the server, identified concept markers and search result documents to generate a ranked list of the identified concept markers and ranked search result documents, wherein:
the search result documents correspond to the documents generated in response to the user query,
the identified concept markers correspond to the concept markers identified, by the concept marker recommender, in the search result documents;
ranking the identified concept markers, wherein ranking the identified concept markers comprises determining a relevance of each identified concept marker to search terms in the user query,
wherein determining the relevance of each identified concept marker comprises generating a semantic vector space comprised of the concept markers;
in response to selection, by a user, of a concept marker from the ranked list of the identified concept markers, re-ranking, by a re-ranker module of the server, ranked concept markers in the ranked list of the identified concept markers and the ranked search result documents to generate re-ranked concept markers and re-ranked documents, wherein re-ranking the ranked concept markers in the ranked list of the identified concept markers and the ranked documents comprises determining a frequency with which the ranked concept markers appear in the ranked documents;
providing the re-ranked concept markers and the re-ranked documents to the user via the user terminal;
gathering, by a training module, metrics associated with system performance; and
adjusting, by the concept marker assigner, assignment of the concept markers based on the metrics.