2021/09/10 by Mayank Soni, Soni, Mayank, Benjamin R. Cowan +3
Computer Science · Psychology · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Mental Health via Writing #Speech and dialogue systems #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2109.05090
openalex publication_date 2021/09/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Neural language modelling has progressed the state-of-the-art in different\ndownstream Natural Language Processing (NLP) tasks. One such area is of\nopen-domain dialog modelling, neural dialog models based on GPT-2 such as\nDialoGPT have shown promising performance in single-turn conversation. However,\nsuch (neural) dialog models have been criticized for generating responses which\nalthough may have relevance to the previous human response, tend to quickly\ndissipate human interest and descend into trivial conversation. One reason for\nsuch performance is the lack of explicit conversation strategy being employed\nin human-machine conversation. Humans employ a range of conversation strategies\nwhile engaging in a conversation, one such key social strategies is\nSelf-disclosure(SD). A phenomenon of revealing information about one-self to\nothers. Social penetration theory (SPT) proposes that communication between two\npeople moves from shallow to deeper levels as the relationship progresses\nprimarily through self-disclosure. Disclosure helps in creating rapport among\nthe participants engaged in a conversation. In this paper, Self-disclosure\nenhancement architecture (SDEA) is introduced utilizing Self-disclosure Topic\nModel (SDTM) during inference stage of a neural dialog model to re-rank\nresponse candidates to enhance self-disclosure in single-turn responses from\nfrom the model.\n