Device and method for machine reading comprehension question and answer
Assignee42 Maru Inc.
InventorsDong Hwan Kim, Hyun Ok Kim, Woo Tae Jeong
Life of this patent
Estimating the term…
Abstract
A machine reading comprehension (MRC) question and answer providing method includes receiving a user question; analyzing the user question; selecting at least one document from at least one domain corresponding to an analyzed user question and searching for a passage, which is a candidate answer determined as being suitable for the user question, in the selected at least one document; obtaining at least one correct answer candidate value by inputting the user question and a corresponding passage into each of at least one MRC question and answer unit; and determining whether the at least one correct answer candidate value is a best answer.
Claims6
A machine reading comprehension (MRC) question and answer providing method comprising:
- receiving a user question data in a speech format or a text format;
- analyzing an intention of the user question data by analyzing a lexical meaning based on an entity name recognized through a morphological analysis of the user question data, extracting a query by restoring an abbreviation word or a substitute word and mapping a plurality of domains corresponding to the query based on a rule-based domain classifier;
- selecting a similar question word candidate group similar to the user question data, selecting a similar question word having a high similarity from the similar question word candidate group by applying word embedding and in case that there is a first answer value corresponding to the similar question word, providing the first answer value to a user;
- wherein in case that there is no first answer value corresponding to the similar question word:
- selecting a plurality of documents corresponding to the query based on the plurality of domains;
- searching for a plurality of passages, sentence by sentence in the plurality of documents and obtaining candidates by applying a plurality of MRC question and answer algorithms to the plurality of passages;
- determining whether there is a second answer value based on reliability values of the candidates or consistency values of the candidates;
- and
- in case that there is the second answer value, providing the second answer value to the user;
- and
- wherein in case that there is no second answer value, providing information indicating no result as an answer to the user.
A machine reading comprehension (MRC) question and answer providing method comprising:
- receiving a user question data;
- analyzing the user question data;
- selecting a plurality of documents from a plurality of domains corresponding to the user question data;
- searching for a plurality of passages including candidates for an answer value determined as being suitable for the user question data, in the plurality of documents;
- obtaining N*M candidates by inputting the user question data and the plurality of passages into a plurality of MRC question and answer units, wherein N is the number of the plurality of passages, and M is the number of the plurality of MRC question and answer units;
- determining the answer value based on whether a ‘reliability value of each of the N*M candidates’ exceeds a ‘threshold value’;
- and
- providing the answer value to a user.
A machine reading comprehension (MRC) question and answer providing method comprising:
- receiving a user question data in a speech format or a text format;
- analyzing the user question data based on a natural language processing analysis and a natural language comprehension analysis;
- selecting a similar question word candidate group similar to the user question data, selecting a similar question word having a high similarity from the similar question word candidate group based on a term frequency-inverse document frequency (TF-IDF) value and in case that there is a first answer value corresponding to the similar question word, providing the first answer value to a user;
- wherein in case that there is no first answer value corresponding to the similar question word:
- selecting a plurality of documents based on a plurality of domain information corresponding to the user question data and extracting a plurality of passages in the plurality of documents;
- obtaining N*M candidates by inputting the user question data and the plurality of passages into a plurality of MRC question and answer units, wherein
- N is the number of the plurality of passages, M is the number of the plurality of MRC question and answer units
- determining whether there is a second answer value based on reliability values of the N*M candidates or consistency values of the N*M candidates, and
- in case that there is the second answer value, providing the second answer value to the user;
- and
- wherein in case that there is no second answer value, providing information indicating no result as an answer to the user.
Dependencies are inferred from the claim text.