Call Centre Conversation Summarization: A Pilot Task at Multiling 2015

Abstract

This paper describes the results of the Call Centre Conversation Summarization task at Multiling'15. The CCCS task consists in generating abstractive synopses from call centre conversations between a caller and an agent. Synopses are summaries of the problem of the caller, and how it is solved by the agent. Generating them is a very challenging task given that deep analysis of the dialogs and text generation are necessary. Three languages were addressed: French, Italian and English translations of conversations from those two languages. The official evaluation metric was ROUGE-2. Two participants submitted a total of four systems which had trouble beating the extractive baselines. The datasets released for the task will allow more research on abstractive dialog summarization.

Beyond Utterance Extraction: Summary Recombination for Speech Summarization

“…We follow the experimental setup of the CCCS shared task [16] except that we have a larger test set. The length limit for synopses is 7% of the conversation words, evaluation is performed with the ROUGE-2 metric.…”

Are We Summarizing the Right Way? A Survey of Dialogue Summarization Data Sets

“…-Abstractive summaries about the main events of the conversations, such as the objective of the caller, whether and how it was solved by the agent, and the attitude of both parties. Synopses written by quality assurance experts from call centres (Favre et al, 2015).…”

Incremental temporal summarization in multi-party meetings

“…Dialogue summarization corpora (Carletta et al, 2005;Janin et al, 2003;Lacson et al, 2006;Favre et al, 2015;Misra et al, 2015;Barker et al, 2016;Liu et al, 2019a;Gliwa et al, 2019) have helped accelerate the research in the area of conversational summarization and helped identify the differences in the dialogue and news article summarization (Jung et al, 2019). Our dataset could help progress the field by identifying similar differences and developing summarization model for incremental scenarios.…”

“…The task of incremental temporal summarization in dialogue that is developed in this work has two main aspects to it, i) The content being summarized has a temporal order, meaning the information evolves over time. All conversations are temporal in nature, however, the current datasets on dialogue summarization (Carletta et al, 2005;Janin et al, 2003;Liu et al, 2019a;Gliwa et al, 2019;Lacson et al, 2006;Favre et al, 2015) consist of summaries that are written for the entire dialogue or parts of it (not in a sequence). Thus the summaries are not in temporal order.…”