US 11,914,954 B2
Methods and systems for generating declarative statements given documents with questions and answers
Edward A. Fox, Blacksburg, VA (US); Saurabh Chakravarty, Christiansburg, VA (US); Raja Venkata Satya Phanindra Chava, Herndon, VA (US); and Maanav Mehrotra, Blacksburg, VA (US)
Assigned to VIRGINIA TECH INTELLECTUAL PROPERTIES, INC., Blacksburg, VA (US)
Filed by VIRGINIA TECH INTELLECTUAL PROPERTIES, INC., Blacksburg, VA (US)
Filed on Dec. 7, 2020, as Appl. No. 17/113,880.
Application 17/113,880 is a continuation of application No. 62/945,202, filed on Dec. 8, 2019.
Prior Publication US 2021/0174016 A1, Jun. 10, 2021
Int. Cl. G06F 40/205 (2020.01); G06N 20/00 (2019.01); G06F 40/279 (2020.01); G06N 3/044 (2023.01); G06Q 50/26 (2012.01); G06Q 30/018 (2023.01)
CPC G06F 40/205 (2020.01) [G06F 40/279 (2020.01); G06N 3/044 (2023.01); G06N 20/00 (2019.01); G06Q 30/018 (2013.01); G06Q 50/265 (2013.01)] 20 Claims
OG exemplary drawing
 
8. A system for transforming question-answer groups into declarative segments, comprising:
a memory device to store computer-readable instructions thereon; and
at least one computing device configured through execution of the computer-readable instructions to:
identify a question-answer group from a plurality of question-answer groups;
parse the question-answer group to identify a question and an answer;
classify the question into a question dialog act category;
classify the answer into an answer dialog act category; and
apply a transformer to the question-answer group to obtain a declarative segment corresponding to the question-answer group,
wherein, to apply the transformer to the question-answer group to obtain the declarative segment corresponding to the question-answer group, the at least one computing device is configured to:
identify a plurality of part-of-speech tags for respective ones of a plurality of words in the question-answer group;
identify a chunk pattern based at least in part on a combination of the question dialog act category and the answer dialog act category;
match the chunk pattern to at least one chunk from the plurality of words based at least in part on the plurality of part-of-speech tags;
identify a transformation rule from a plurality of
transformation rules based at least in part on the chunk pattern and
the combination of the question dialog act category and the answer dialog act category;
apply the transformation rule to the at least one chunk to obtain at least one transformed chunk; and
generate the declarative segment based at least in part on the at least one transformed chunk.