Reproduce “A Survey of Personality, Persona, and Profile in Conversational Agents and Chatbots” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Richard Sutcliffe
Richard Sutcliffe (University of Essex) provides a systematic review of personality, persona, and profile in neural conversational agents, clarifying terminology and categorizing 21 datasets, various personality schemes, and embodiment methods used from 2016 onwards. The work identifies significant challenges in the consistent evaluation of personality in these agents.
Key insights
- The terms "Personality," "Persona," and "Profile" are frequently used interchangeably in CA research, necessitating a flexible analytical approach despite their distinct conceptual origins.
- Personality embodiment in CAs predominantly relies on neural architectures like Seq-to-Seq and Transformers, utilizing methods such as conditioning vectors, memory networks, and descriptive input prefixing.
- The field faces significant challenges in developing robust and consistent evaluation methodologies for assessing personality in conversational agents, leading to inconclusive results across many studies.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.