Ubuntu-fr: A Large and Open Corpus for Multi-modal Analysis of Online Written Conversations

Ubuntu-fr: a Large and Open Corpus for Supporting Multi-Modality and Online Written Conversation Studies Nicolas Hernandez, Soufian Salim, Elizaveta Loginova Clouet LINA UMR 6241 - Université de Nantes 2 rue de la houssinière, 44322 Nantes Cedex 03 {firstname}.{lastname}@univ-nantes.fr Abstract We present a large, free, French corpus of online written conversations extracted from the Ubuntu platform’s forums, mailing lists and IRC channels. The corpus is meant to support multi-modality and diachronic studies of online written conversations. We choose to build the corpus around a robust metadata model based upon strong principles, such as the "stand off" annotation principle. We detail the model, we explain how the data was collected and processed - in terms of meta-data, text and conversation - and we detail the corpus’ contents through a series of meaningful statistics. A portion of the corpus - about 4,700 sentences from emails, forum posts and chat messages sent in November 2014 - is annotated in terms of dialogue acts and sentiment. We discuss how we adapted our dialogue act taxonomy from the DIT++ annotation scheme and how the data was annotated, before presenting our results as well as a brief qualitative analysis of the annotated data. Keywords: corpus, conversation, dialogue, chat, forum, email 1. Introduction Help desk and customer service are the most common ar- eas of Customer Relationship Management (CRM) tech- nology investment1. In order to offer new capabilities to CRM software, the ODISAE2 project aims at developing a semantic analyser of written online conversations across several modalities (i.e. chat, forum, email). These capabili- ties are: multi-modal text information retrieval (e.g. finding the solution to a problem in a modality different from the one in which the request was formulated), automated FAQ and documentation management (e.g. automatic detection of the absence of a suitable solution to a recurring request), automated assistance generation (e.g. helping users to for- mulate problems, evaluating answers’ exhaustivity), or con- versation supervision (e.g. detecting attrition, irritation). The project presents an opportunity to make the academic and the industrial worlds collaborate on real-world use- cases. Unfortunately, it also comes with heavy restrictions when it comes to disseminating data and results of scien- tific analysis. We consider that making our research repro- ducible and reusable in an open science perspective is an important issue (Nielsen, 2011). In this paper, we present our efforts to build a large, free, French corpus of online written conversations from syn- chronous and asynchronous mediums and assorted explana- tory texts. All items were gathered on the same period and pertain to the same type of discursive situations, namely user assistance in problem solving tasks. Our object of study is a specific domain of computer-mediated communi- cation (CMC), computer-mediated conversations, and our scientific objectives are multiple: 1. Reaching a better understanding of the structure of problem-solving conversations; 1Software Advice survey 2014 "CRM software users" www. softwareadvice.com/crm/userview/report-2014 2ODISAE is supported by the Unique Inter-ministerial Fund (FUI) no. 17 (wwww.odisae.com) 2. Exploiting modality comparability in order to improve linguistic processing of less conventional user gen- …

Content truncated to 500 words