SYSTEM AND METHOD FOR LANGUAGE PROCESSING
A system and a method for language processing including a computing device; a memory; and a processor, the processor configured to: identify a question and context for the question in an interaction recording; identify whether the context for the question includes an answer to the question using a machine learning mode, and: when the context for the question includes the answer to the question, generate a question-answer combination and store the combination in a database; and when the context does not include the answer for the question, store the question in the database.
1 . A method of language processing comprising:
identifying a question and context for the question in an interaction recording;
identifying whether the context for the question comprises an answer to the question using a machine learning model, and:
when the context for the question includes the answer to the question, generating a question-answer combination and storing the combination in a database; and
when the context does not include the answer for the question, storing the question in the database.
2 . A method according to claim 1 , wherein the method comprises separating the question-answer combination from the interaction recording using the machine learning model.
3 . A method according to claim 1 , wherein the machine learning model is a large language model.
4 . A method according to claim 1 , comprising querying the database for a recorded question-answer combination that comprises the question-answer combination, and:
when the recorded question-answer combination comprises the question of the question-answer combination, increasing a counter for the recorded question and assessing whether the answer of the recorded question-answer combination comprises the answer of the recorded question-answer combination:
when the recorded question-answer combination comprises the answer of the question-answer combination, increasing a counter for the recorded answer; and
when the recorded question-answer combination does not comprise the answer of the question-answer combination, adding the answer to the recorded question; and
when the recorded question-answer combination does not comprise the question of the question-answer combination, adding the question-answer combination to the database.
5 . A method according to claim 4 , wherein the method comprises periodically identifying question-answer combinations which have a counter value that is above a threshold counter value and generating a list of frequently asked questions and answers to the questions-answer combinations.
6 . A method according to claim 5 , comprising using the list of frequently asked questions and answers in the automatic provision of an answer to a received question.
7 . A method according to claim 5 , comprising using the list of frequently asked questions and answers in providing an agent with recommended answers to a customer query.
8 . A method according to claim 4 , wherein a similarity between the question and the question in the recorded question-answer combination is identified by assessing a semantic similarity between the questions.
9 . A method according to claim 8 , wherein the semantic similarity between the questions is assessed by vectorial representation of the question and the recorded question and calculation of a cosine similarity.
10 . A method according to claim 1 , comprising automatically updating recorded question-answer combinations by updating one or more of: a question, an answer to a question, a question-answer combination.
11 . A system for language processing, the system comprising:
a computing device;
a memory; and
a processor, the processor configured to:
identify a question and context for the question in an interaction recording;
identify whether the context for the question comprises an answer to the question using a machine learning mode, and:
when the context for the question includes the answer to the question, generate a question-answer combination and store the combination in a database; and
when the context does not include the answer for the question, store the question in the database.
12 . A system according to claim 11 , wherein the processor is configured to separate the question-answer combination from the interaction recording using the machine learning model.
13 . A system according to claim 11 , wherein the machine learning model is a large language model.
14 . A system according to claim 11 , wherein the processor is configured to query the database for a recorded question-answer combination that comprises the question-answer combination, and:
when the recorded question-answer combination comprises the question of the question-answer combination, increasing a counter for the recorded question and assessing whether the answer of the recorded question-answer combination comprises the answer of the recorded question-answer combination:
when the recorded question-answer combination comprises the answer of the question-answer combination, increasing a counter for the recorded answer; and
when the recorded question-answer combination does not comprise the answer of the question-answer combination, adding the answer to the recorded question;
and
when the recorded question-answer combination does not comprise the question of the question-answer combination, adding the question-answer combination to the database.
15 . A system according to claim 14 , wherein the processor is configured to periodically identify question-answer combinations which have a counter value that is above a threshold counter value and generating a list of frequently asked questions and answers to the questions-answer combinations.
16 . A system according to claim 15 , wherein the processor is configured to use the list of frequently asked questions and answers in the automatic provision of an answer to a received question.
17 . A system according to claim 15 , wherein the processor is configured to use the list of frequently asked questions and answers in providing an agent with recommended answers to a customer query.
18 . A system according to claim 14 , wherein a similarity between the question and the question in the recorded question-answer combination is identified by assessing a semantic similarity between the questions.
19 . A system according to claim 18 , wherein the semantic similarity between the questions is assessed by vectorial representation of the question and the recorded question and calculation of a cosine similarity.
20 . A method of language processing of interaction recordings, the method comprising:
identifying a question and context for the question in a recording of an interaction;
identifying whether the context for the question comprises a response to the question using a machine learning model:
when the context for the question includes the response to the question, creating a question-response pair and storing the pair in a database; and
when the context does not include the response for the question, storing the question in the database.