Abstract generation device, method, program, and recording medium

1. A summary generating apparatus comprising a processor configured to execute operations comprising:

converting an input utterance sequence into a confusion network sequence constituted by a k-best of candidate words of speech recognition results;

generating, a lattice sequence having the candidate words as internal nodes and a combination of k words among the candidate words for an identical speech as an external node, in which edges are extended between internal nodes other than internal nodes included in an identical external node, from the confusion network sequence;

generating an integer programming problem for selecting a path that maximizes an objective function including at least a coverage score of an important word, of paths following the internal nodes with the edges extended, in the lattice sequence; and

generating a summary of the utterance sequence using the candidate words indicated by the internal nodes included in the path selected by solving the integer programming problem, under a constraint on a length of a summary to be generated.

2. The summary generating apparatus according to claim 1 , wherein the coverage score of the important word is a score that increases when the number of candidate words included in the summary to be generated increases, among candidate words that are independent words included in the lattice sequence.

3. A computer-readable non-transitory recording medium storing computer-executable summary generating program instructions that when executed by a processor cause a computer system to execute operations as the summary generating apparatus according to claim 2 .

4. The summary generating apparatus according to claim 2 , wherein the objective function further comprises a score of the internal node represented by an importance of a candidate word included in the summary to be generated, and a score of an edge indicating a likelihood of connection between candidate words at both ends of the edge included in the summary to be generated.

5. A computer-readable non-transitory recording medium storing computer-executable summary generating program instructions that when executed by a processor cause a computer system to execute operations as the summary generating apparatus according to claim 4 .

6. The summary generating apparatus according to claim 1 , wherein the objective function further comprises a score of the internal node represented by an importance of a candidate word included in the summary to be generated, and a score of an edge indicating a likelihood of connection between candidate words at both ends of the edge included in the summary to be generated.

7. The summary generating apparatus according to claim 6 , wherein the score of the internal node comprises an appearance frequency and an inverse document frequency of the candidate word and a confidence of speech recognition for the candidate word.

8. A computer-readable non-transitory recording medium storing computer-executable summary generating program instructions that when executed by a processor cause a computer system to execute operations as the summary generating apparatus according to claim 7 .

9. The summary generating apparatus according to claim 6 , wherein the score of the edge comprises a bigram appearance rate of candidate words at both ends of the edge.

10. A computer-readable non-transitory recording medium storing computer-executable summary generating program instructions that when executed by a processor cause a computer system to execute operations as f the summary generating apparatus according to claim 9 .

11. A computer-readable non-transitory recording medium storing computer-executable summary generating program instructions that when executed by a processor cause a computer system to execute operations as the summary generating apparatus according to claim 6 .

12. A computer-readable non-transitory recording medium storing computer-executable summary generating program instructions that when executed by a processor cause a computer system to execute operations as the summary generating apparatus according to claim 1 .

13. A computer-readable non-transitory storage medium that stores a summary generating program that when executed by a processor cause a computer to execute operations as the summary generating apparatus according to claim 1 .

14. A summary generating method performed at a summary generating apparatus comprising a speech recognition unit, a lattice generating unit, an integer programming problem generating unit, and a summary generating unit, the method comprising:

converting, at the speech recognition unit, an input utterance sequence into a confusion network sequence constituted by a k-best of candidate words of speech recognition results;

generating, at the lattice generating unit, a lattice sequence having the candidate words as internal nodes and a combination of k words among the candidate words for an identical speech as an external node, in which edges are extended between internal nodes other than internal nodes included in an identical external node, from the confusion network sequence;

generating, at the integer programming problem generating unit, an integer programming problem for selecting a path that maximizes an objective function including at least a coverage score of an important word, of paths following the internal nodes with the edges extended, in the lattice sequence; and

generating, at the summary generating unit, a summary of the utterance sequence using the candidate words indicated by the internal nodes included in the path selected by solving the integer programming problem, under a constraint on a length of a summary to be generated.