Exploring the context of recurrent neural network based conversational agents
Raffaele Piccini and Gerasimos Spanakis


Conversational agents have begun to rise both in the academic (in terms of research) and commercial (in terms of applications) world. This paper investigates the task of building a non-goal driven conversational agent, using neural network generative models and analyzes how the conversation context is handled. It compares a simpler Encoder-Decoder with a Hierarchical Recurrent Encoder-Decoder architecture, which includes an additional module to model the context of the conversation using previous utterances information.

We found that the hierarchical model was able to extract relevant context information and include them in the generation of the output. However, it performed worse than the simple Encoder-Decoder model regarding both grammatically correct output and meaningful response. Despite these results, experiments demonstrate how conversations about similar topics appear close to each other in the context space due to the increased frequency of specific topic-related words. We attempted a classification of the context space to the relevant topics of the conversation and showed that while it's easy to predict more tailored topics (like tourism), it is really challenging to predict that the topic of a conversation is about general topics like every day life. These interesting results leave promising directions for future research and how the context of a conversation can be exploited.