Bibliographic record
Abstract
Mental Model Ascription by Language-Enabled Intelligent Agents Marjorie McShane (marge@umbc.edu) Department of Computer Science and Electrical Engineering University of Maryland Baltimore County Baltimore, MD, 21250, USA The main goal of the workshop is to foster mutual learning, discussion and future collaboration among researchers pursuing agent-oriented mental model ascription in integrative cognitive architectures. Topic and Goal Mental model ascription can be defined as inferring features of another human or artificial agent that cannot be directly observed, such as that agent’s beliefs, plans, goals, intentions, personality traits, mental and emotional states, and knowledge about the world. This capability is an essential functionality of intelligent agents if they are to engage in sophisticated collaborations with people. The computational modeling of mental model ascription offers an excellent opportunity to explore the interaction of traditionally separate modules of cognitive architectures, such as language understanding, plan- and goal-oriented reasoning, and memory management. The study of mental model ascription can benefit from advances in fields as disparate as machine reasoning, social interaction, developmental psychology, robotics, emotion, philosophy and computational linguistics, to name just a few. 1 The common thread of this workshop will be the computational modeling of unobservable features by intelligent agents using language input as at least one of their modes of perception. Topics of interest include but are not limited to: Program Committee (confirmed) Ron Artstein (USC) Jerry Ball (Air Force Research Laboratory) Paul Bello (Office of Naval Research) Graeme Hirst (University of Toronto) Eva Hudlicka (Psychometrix Associates, Inc.) Pat Langley (University of Auckland, NZ and CMU) Marjorie McShane, Chair (UMBC) Sergei Nirenburg (UMBC) Massimo Poesio (University of Essex) Chris Potts (Stanford University) Yorick Wilks (IHMC). Organizational This will be a full day workshop that will include invited talks, talks selected by abstract submission, a round table discussion, and, optionally, a poster session. Talks will be grouped by similarity of theme and approach, and the schedule will allow for extended discussion of each group of presentations, best exploiting the workshop genre. We expect 30-40 participants that include students and researchers with broad interests in the computational modeling of cognition and/or psychologically-inspired natural language processing. The final session of the day will be devoted to planning a special journal issue (for Advances in Cognitive Systems) of papers inspired by the workshop. There are no special requirements for participants in the workshop. The workshop website is http://ilit.umbc.edu/Workshop/MentalModelCogSci2013.html . The contact email is mentalmodel2013@gmail.com. The workshop organizer, Marjorie McShane, has been working in the field of AI-NLP for the past fifteen years, with recent work focusing on the development of cognitive simulations of virtual patients to support clinician training. For a brief CV and list of publications, see http://ilit.umbc.edu/PubMcShane.htm. Developing computational treatments of language phenomena (e.g., indirect speech acts, irony, paraphrase, humor, coercion) that require or give rise to mental model ascription. Applying computational models of other cognitive capabilities (dialog, emotion, agent collaboration/competition and plan- and goal- oriented reasoning) to mental model ascription. Modeling agent decisions about what to learn about other agents’ unobservable features, considering that attempting to learn everything in every context would incur a heavy cognitive load. Modeling how agents measure their confidence in the results of mental model ascription, which will be affected by their confidence in their understanding of contributing linguistic (or other) percepts as well as their ability to make valid inferences. Modeling dynamic belief modification, including overriding a previous belief and managing memories with respect to modified beliefs. As a comparison, CogSci 2012 featured a workshop, Modeling the Perception of Intention that treated intention recognition with an emphasis on visual perception.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.012 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.002 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.004 | 0.006 |
| Open science | 0.001 | 0.004 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".