US 12,086,174 B2
Sentence data analysis information generation device using ontology, sentence data analysis information generation method, and sentence data analysis information generation program
Jingyu Sun, Musashino (JP); and Susumu Takeuchi, Musashino (JP)
Assigned to Nippon Telegraph and Telephone Corporation, Tokyo (JP)
Appl. No. 17/912,236
Filed by Nippon Telegraph and Telephone Corporation, Tokyo (JP)
PCT Filed Apr. 10, 2020, PCT No. PCT/JP2020/016095
§ 371(c)(1), (2) Date Sep. 16, 2022,
PCT Pub. No. WO2021/205639, PCT Pub. Date Oct. 14, 2021.
Prior Publication US 2023/0140938 A1, May 11, 2023
Int. Cl. G06F 16/36 (2019.01); G06F 40/137 (2020.01); G06F 40/205 (2020.01)
CPC G06F 16/367 (2019.01) [G06F 40/137 (2020.01); G06F 40/205 (2020.01)] 12 Claims
OG exemplary drawing
 
1. An ontology-based text data analysis information generation device comprising:
a dependence relationship tree information generation unit, including one or more processors, configured to generate dependence relationship tree information that indicates, in a tree structure, a dependence relationship between words in text data to be analyzed;
a graph information generation unit, including one or more processors, configured to extract a subject, a predicate, and an object from the text data based on the dependence relationship tree information, and generate graph information that indicates, in a graph structure, triple information of ontology comprising the extracted subject, predicate, and object,
wherein the triple information of ontology includes a main triple information and a secondary triple information of the text data,
wherein extracting the predicate comprises: (i) extracting a verb word based on a root of the dependency relationship tree information as a predicate of the main triple information, and (ii) extracting a modified word that depends on the word of the root as a predicate of the secondary triple information, and
wherein extracting the predicate of the main triple information comprises:
determining whether the word in the root is a verb word,
in response to determining that the word in the root is not a verb word, adding a modified word that depends on the word of the root to the word in the root to obtain a combined verb word, and
exacting the combined verb word as the predicate of the main triple information; and
a hierarchical concept information adding unit, including one or more processors, configured to extract two suitable components from the text data, acquire broader concept information that is common between the two components, and add the acquired broader concept information to the graph information as a parent node that is common between the two components so as to generate text data analysis information.