Email Summarizer and Action Item Extractor

Disciplines

Artificial Intelligence and Robotics | Computer Sciences

Abstract (300 words maximum)

Countless emails are sent and received every day, and a lot of time is spent reading through and understanding the content of these emails. This project aims to increase the efficiency of reading and gathering relevant information from emails. Our solution includes two parts: abstractive text summarization and action item extraction. Currently, these two items are common in different domains, however, they have not been combined and used with email understanding. Abstractive text summarization is the process of outputting the ideas of the emails using different words without giving direct sentences from the document. We do this by taking an input of at most 6000 words from the chain of emails, and then we output 700 words for the summarization. In this way, a person would be able to tell if the email is something they will need to look at, and if not, then the Action Item Extractor tells what actions need to be performed Action item extraction involves finding all instances in a piece of text that are instructions, dates, or require something from the recipient. For example, “Please give me a list of names who are attending the party”, “Send the signed document to me by Friday '', and “Don’t forget to sign-up for the costume contest”. Dates and deadlines are also extracted from the text. These two solutions hope to improve the efficiency of parsing through emails.

Academic department under which the project should be listed

CCSE - Computer Science

Primary Investigator (PI) Name

Md Abdullah Al Hafiz Khan

This document is currently not available here.

Share

COinS
 

Email Summarizer and Action Item Extractor

Countless emails are sent and received every day, and a lot of time is spent reading through and understanding the content of these emails. This project aims to increase the efficiency of reading and gathering relevant information from emails. Our solution includes two parts: abstractive text summarization and action item extraction. Currently, these two items are common in different domains, however, they have not been combined and used with email understanding. Abstractive text summarization is the process of outputting the ideas of the emails using different words without giving direct sentences from the document. We do this by taking an input of at most 6000 words from the chain of emails, and then we output 700 words for the summarization. In this way, a person would be able to tell if the email is something they will need to look at, and if not, then the Action Item Extractor tells what actions need to be performed Action item extraction involves finding all instances in a piece of text that are instructions, dates, or require something from the recipient. For example, “Please give me a list of names who are attending the party”, “Send the signed document to me by Friday '', and “Don’t forget to sign-up for the costume contest”. Dates and deadlines are also extracted from the text. These two solutions hope to improve the efficiency of parsing through emails.