Krishna Asawa

Akanksha Mehndiratta, Krishna Asawa

Multiturn dialogue models aim to generate human-like responses by leveraging conversational context, consisting of utterances from previous exchanges. Existing methods often neglect the interactions between these utterances or treat all of them as equally significant. This paper introduces a discourse-aware framework for response selection in retrieval-based dialogue systems. The proposed model first encodes each utterance and response with contextual, positional, and syntactic features using Multi-view Canonical Correlation Analysis (MCCA). It then learns discourse tokens that capture relationships between an utterance and its surrounding turns in a shared subspace via Canonical Correlation Analysis (CCA). This two-step approach effectively integrates semantic and syntactic features to build discourse-level understanding. Experiments on the Ubuntu Dialogue Corpus demonstrate that our model achieves significant improvements in automatic evaluation metrics, highlighting its effectiveness in response selection.

Akanksha Mehndiratta, Krishna Asawa

The evolution of conversational agents has been driven by the need for more contextually aware systems that can effectively manage dialogue over extended interactions. To address the limitations of existing models in capturing and utilizing long-term conversational history, we propose a novel framework that integrates Deep Canonical Correlation Analysis (DCCA) for discourse-level understanding. This framework learns discourse tokens to capture relationships between utterances and their surrounding context, enabling a better understanding of long-term dependencies. Experiments on the Ubuntu Dialogue Corpus demonstrate significant enhancement in response selection, based on the improved automatic evaluation metric scores. The results highlight the potential of DCCA in improving dialogue systems by allowing them to filter out irrelevant context and retain critical discourse information for more accurate response retrieval.

Akanksha Mehndiratta, Krishna Asawa

Canonical Correlation Analysis (CCA) has been exploited immensely for learning latent representations in various fields. This study takes a step further by demonstrating the potential of CCA in identifying Elementary Discourse Units(EDUs) that captures the latent information within the textual data. The probabilistic interpretation of CCA discussed in this study utilizes the two-view nature of textual data, i.e. the consecutive sentences in a document or turns in a dyadic conversation, and has a strong theoretical foundation. Furthermore, this study proposes a model for Elementary Discourse Unit(EDU) segmentation that discovers EDUs in textual data without any supervision. To validate the model, the EDUs are utilized as textual unit for content selection in textual similarity task. Empirical results on Semantic Textual Similarity(STSB) and Mohler datasets confirm that, despite represented as a unigram, the EDUs deliver competitive results and can even beat various sophisticated supervised techniques. The model is simple, linear, adaptable and language independent making it an ideal baseline particularly when labeled training data is scarce or nonexistent.