ScenarioSA: A Dyadic Conversational Database for Interactive Sentiment Analysis

Interactive sentiment analysis is an emerging, yet challenging, subtask of the natural language processing problem. It aims to discover the affective state and sentimental change of each person in a conversation, and has attracted an increasing attention from both academia and industry. Existing sentiment analysis approaches are insufficient in modelling the interactions among people. However, the development of new approaches are critically limited by the lack of labelled interactive sentiment datasets. In this paper, we present a new conversational database that we have created and made publicly available, namely ScenarioSA, for interactive sentiment analysis. We manually label 2,214 multi-turn English conversations collected from various websites that provide online communication services. In comparison with existing sentiment datasets, ScenarioSA (1) is no longer limited to one specific domain but covers a wide range of topics and scenarios; (2) describes the interactions between two speakers of each conversation; and (3) reflects the sentimental evolution of each speaker over the course of a conversation. Finally, we propose an extension of interactive attention networks that could model the interactions, and compare various strong sentiment analysis algorithms on ScenarioSA, demonstrating the need of novel interactive sentiment analysis models and the potential of ScenarioSA to facilitate the development of such models.