Performing linguistic analysis by scoring syntactic graphs

• US 10,360,305 B2
• Filed: 02/16/2016
• Issued: 07/23/2019
• Est. Priority Date: 03/26/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: VirtuOz SA (Microsoft Corporation)
• Inventors: Larcheveque, Jean-Marie Henri Daniel, Powers, Elizabeth Ireland, Recksiek, Freya Kate, Teodosiu, Dan
• Primary Examiner(s): Ortiz-Sanchez, Michael
• Application Number:
  US15/044,977
• Publication Number:
  US 20160239480A1
• Time in Patent Office: 1,253 Days
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning

Abstract

Semantic clustering techniques are described. In various implementations, a conversational agent is configured to perform semantic clustering of a corpus of user utterances. Semantic clustering may be used to provide a variety of functionality, such as to group a corpus of utterances into semantic clusters in which each cluster pertains to a similar topic. These clusters may then be leveraged to identify topics and assess their relative importance, as for example to prioritize topics whose handling by the conversation agent should be improved. A variety of utterances may be processed using these techniques, such as spoken words, textual descriptions entered via live chat, instant messaging, a website interface, email, SMS, a social network, a blogging or micro-blogging interface, and so on.

Claims

1. A computer-implemented method of improving performance of linguistic processing by an automated conversational agent, the method comprising:

mapping sequences of one or more user utterances to lemmata, the one or more user utterances containing variations in lexical form, thereby yielding mapped lemmata;

unifying constructions of the mapped lemmata to generate one or more syntactic graphs;

scoring each of the one or more syntactic graphs based on confidence in the syntactic graph'"'"'s recognition of particular constructions of the lemmata, the particular constructions including syntactic patterns linking a given lemma to one or more other lexical items;

using a given syntactic graph with a highest determined score in an abstracted semantic graph of the one or more user utterances;

providing the abstracted semantic graph to the automated conversational agent in a format that enables the automated conversational agent to employ the abstracted semantic graph; and

participating, by the automated conversational agent, in an interactive natural language dialog with a user by the employed abstracted semantic graph recognizing and processing user utterances having variations in lexical form.

2. The method as claimed in claim 1, wherein the sequences comprise word sequences segmented from the one or more user utterances.

3. The method as claimed in claim 1, wherein mapping a sequence comprises matching the sequence to a flexion that includes a respective lemma.

4. The method as claimed in claim 3, wherein a lexicon is used to match the sequence to the flexion including the respective lemma.

5. The method as claimed in claim 1, wherein the unified constructions include part-of-speech information and syntactic pattern linking information.

6. The method as claimed in claim 1, wherein a measure of confidence is assigned to the recognizableness of the particular constructions, and determining the score is based on the assigned measure of confidence.

7. The method as claimed in claim 1, wherein determining the score is further based on linearity in occurrences of objects after verbs in the respective unified constructions.

8. The method as claimed in claim 1, wherein determining the score is further based on part-of-speech information of the respective unified constructions.

9. The method as claimed in claim 1, wherein the mapping of the given syntactic graph with a highest score enables abstraction of the semantic graph in representing logical relations between ontology concepts.

10. A computer system for improving performance of linguistic processing by an automated conversational agent, the system comprising:

one or more computer processors operatively coupled to computer memory, the one or more processors configured to;

map sequences of one or more user utterances to lemmata, the one or more user utterances containing variations in lexical form, thereby yielding mapped lemmata;

unify constructions of the mapped lemmata to generate one or more syntactic graphs;

score each of the one or more syntactic graphs based on confidence in the syntactic graph'"'"'s recognition of particular constructions of the lemmata, the particular constructions including syntactic patterns linking a given lemma to one or more other lexical items;

use a given syntactic graph with a highest determined score in an abstracted semantic graph of the one or more user utterances; and

provide the abstracted semantic graph to the automated conversational agent in a format that enables the automated conversational agent to employ the abstracted semantic graph; and

participate, by the automated conversational agent, in an interactive natural language dialog with a user by the employed abstracted semantic graph recognizing and processing user utterances having variations in lexical form.

11. The system as claimed in claim 10, wherein the sequences comprise word sequences segmented from the one or more user utterances.

12. The system as claimed in claim 10, wherein the one or more processors are further configured to map a sequence by matching the sequence to a flexion that includes a respective lemma.

13. The system as claimed in claim 12, wherein the one or more processors are further configured to use a lexicon to match the sequence to the flexion that includes the respective lemma.

14. The system as claimed in claim 10, wherein the unified constructions include part-of-speech information and syntactic pattern linking information.

15. The system as claimed in claim 10, wherein the one or more processors are further configured to assign a measure of confidence to the recognizableness of the particular constructions, and determines the score based on the assigned measure of confidence.

16. The system as claimed in claim 10, wherein the one or more processors are further configured to determine linearity based on objects occurring after verbs in the respective unified constructions.

17. The system as claimed in claim 10, wherein the one or more processors are further configured to determine the score based on part-of-speech information of the respective unified constructions.

18. The system as claimed in claim 10, wherein the mapping of the given syntactic graph with the highest determined score enables abstraction of the semantic graph in representing logical relations between ontology concepts.

19. One or more non-transitory computer-readable media comprising instructions stored thereon that, responsive to execution by a computing device, causes the computing device to improve performance of linguistic processing by an automated conversational agent by:

mapping sequences of one or more user utterances to lemmata, the one or more user utterances containing variations in lexical form, thereby yielding mapped lemmata;

unifying constructions of the mapped lemmata to generate one or more syntactic graphs;

scoring each of the one or more syntactic graphs based on confidence in the syntactic graph'"'"'s recognition of particular constructions of the lemmata, the particular constructions including syntactic patterns linking a given lemma to one or more other lexical items;

using a given syntactic graph with a highest determined score in an abstracted semantic graph of the one or more user utterances;

providing the abstracted semantic graph to the automated conversational agent in a format that enables the automated conversational agent to employ the abstracted semantic graph; and

participating, by the automated conversational agent, in an interactive natural language dialog with a user by the employed abstracted semantic graph recognizing and processing user utterances having variations in lexical form.