Automatic measurement of semantic similarity of conversations

Patent metadata

Field	Value
Publication number	US-11823666-B2
Application number	US-202117492716-A
Country	US
Kind code	B2
Filing date	Oct 4, 2021
Priority date	Oct 4, 2021
Publication date	Nov 21, 2023
Grant date	Nov 21, 2023

Abstract

Automatic measurement of semantic textual similarity of conversations, by: receiving two conversation texts, each comprising a sequence of utterances; encoding each of the sequences of utterances into a corresponding sequence of semantic representations; computing a minimal edit distance between the sequences of semantic representations; and, based on the computation of the minimal edit distance, performing at least one of: quantifying a semantic similarity between the two conversation texts, and outputting an alignment of the two sequences of utterances with each other.

First claim

What is claimed is:

1. A computer-implemented method comprising:
receiving two conversation texts, each comprising a sequence of utterances;
encoding each of the sequences of utterances, using a machine learning algorithm for semantic encoding of texts, into a corresponding sequence of semantic representations;
computing a minimal edit distance between the sequences of semantic representations, wherein the computing comprises:
assignment of costs to the following edit operations: deletion, insertion, and substitution, wherein the cost of substitution is based on a cosine distance between the semantic representations, and
assigning an infinitely high cost of substitution between those of the semantic representations whose underlying utterances were authored by different author types; and

based on the computation of the minimal edit distance, performing at least one of:
quantifying a semantic similarity between the two conversation texts, and
outputting an alignment of the two sequences of utterances with each other.

2. The computer-implemented method of claim 1 , wherein the semantic representations are semantic distributional representations.

3. The computer-implemented method of claim 1 , executed by at least one hardware processor.

4. A system comprising:
(a) at least one hardware processor; and
(b) a non-transitory computer-readable storage medium having program code embodied therewith, the program code executable by said at least one hardware processor to, automatically:
receive two conversation texts, each comprising a sequence of utterances,
encode each of the sequences of utterances, using a machine learning algorithm for semantic encoding of texts, into a corresponding sequence of semantic representations,
compute a minimal edit distance between the sequences of semantic representations, wherein the computing comprises:
assignment of costs to the following edit operations: deletion, insertion, and substitution, wherein the cost of substitution is based on a cosine distance between the semantic presentations, and
assigning an infinitely high cost of substitution between those of the semantic representations whose underlying utterances were authored by different author types, and

based on the computation of the minimal edit distance, perform at least one of:
quantify a semantic similarity between the two conversation texts, and
output an alignment of the two sequences of utterances with each other.

5. The system of claim 4 , wherein the semantic representations are semantic distributional representations.

6. A computer program product comprising a non-transitory computer-readable storage medium having program code embodied therewith, the program code executable by at least one hardware processor to, automatically:
receiving two conversation texts, each comprising a sequence of utterances;
encoding each of the sequences of utterances, using a machine learning algorithm for semantic encoding of texts, into a corresponding sequence of semantic representations;
computing a minimal edit distance between the sequences of semantic representations, wherein the computing comprises:
assignment of costs to the following edit operations: deletion, insertion, and substitution, wherein the cost of substitution is based on a cosine distance between the semantic representations, and
assigning an infinitely high cost of substitution between those of the semantic representations whose underlying utterances were authored by different author types; and

based on the computation of the minimal edit distance, performing at least one of:
quantifying a semantic similarity between the two conversation texts, and
outputting an alignment of the two sequences of utterances with each other.

7. The computer program product of claim 6 , wherein the semantic representations are semantic distributional representations.

Assignees

• IBM

Inventors

• Lavi Ofer
• Ronen Inbal
• Rabinovich Ella
• Boaz David
• Amid David
• Shlomov Segev
• Anaby - Tavor Ateret

Classifications

• G06F40/289

  Phrasal analysis, e.g. finite state techniques or chunking · CPC title

• G06F40/284

  Lexical analysis, e.g. tokenisation or collocates · CPC title

• G06F40/216

  using statistical methods · CPC title

• G06F40/194

  Calculation of difference between files · CPC title

• G06F40/30 Primary

  Semantic analysis · CPC title