Generating answers from concept-based representation of a topic oriented pipeline

A computer system for generating an answer in a topic oriented pipeline answering system, comprising:

• one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage medium, and program instructions stored on at least one of the one or more tangible storage medium for execution by at least one of the one or more processors via at least one of the one or more memories, wherein the computer system is capable of performing a method comprising:
• receiving a question;
• identifying a candidate answer from a plurality of topics stored in a topic database, wherein the plurality of topics is derived from a corpus of textual documents;
• determining a plurality of sentences based on the identified candidate answer;
• generating a subgraph based on the plurality of sentences and the candidate answer, wherein the subgraph has a plurality of edges and a plurality of nodes corresponding to the plurality of sentences and the candidate answer;
• adding a first node to the subgraph corresponding to the received question, wherein adding the first node includes adding a plurality of question edges to the plurality of edges, and wherein the added plurality of question edges span between the added first node and each node within the plurality of nodes of the generated subgraph;
• calculating a similarity value for each sentence within the plurality of sentences based on comparing the plurality of sentences to the candidate answer and the received question, wherein the calculated similarity value incorporates a distance between the plurality of sentences within the corpus of textual documents that the plurality of sentences was derived from;
• assigning an edge weight to each edge within the plurality of edges based on the calculated similarity value for each sentence within the plurality of sentences;
• identifying at least one sentence within the plurality of sentences with the assigned edge weight that exceeds a threshold value by traversing the generated subgraph;
• generating the answer, wherein the generated answer comprises the candidate answer and the identified at least one sentence;
• determining a score for the generated answer based on an amount of supporting evidence found by searching the corpus;
• and
• outputting the generated answer and the determined score.

A computer program product for generating an answer in a topic oriented pipeline answering system, comprising:

• one or more computer-readable storage medium and program instructions stored on at least one of the one or more tangible storage medium, the program instructions executable by a processor, the program instructions comprising:
• program instructions to receive a question;
• program instructions to identify a candidate answer from a plurality of topics stored in a topic database, wherein the plurality of topics is derived from a corpus of textual documents;
• program instructions to determine a plurality of sentences based on the identified candidate answer;
• program instructions to generate a subgraph based on the plurality of sentences and the candidate answer, wherein the subgraph has a plurality of edges and a plurality of nodes corresponding to the plurality of sentences and the candidate answer;
• program instructions to add a first node to the subgraph corresponding to the received question, wherein adding the first node includes adding a plurality of question edges to the plurality of edges, and wherein the added plurality of question edges span between the added first node and each node within the plurality of nodes of the generated subgraph;
• program instructions to calculate a similarity value for each sentence within the plurality of sentences based on comparing the plurality of sentences to the candidate answer and the received question, wherein the calculated similarity value incorporates a distance between the plurality of sentences within the corpus of textual documents that the plurality of sentences was derived from;
• program instructions to assign an edge weight to each edge within the plurality of edges based on the calculated similarity value for each sentence within the plurality of sentences;
• program instructions to identify at least one sentence within the plurality of sentences with the assigned edge weight that exceeds a threshold value by traversing the generated subgraph;
• program instructions to generate the answer, wherein the generated answer comprises the candidate answer and the identified at least one sentence;
• program instructions to determine a score for the generated answer based on an amount of supporting evidence found by searching the corpus;
• and
• program instructions to output the generated answer and the determined score.