Analogy-ANGLE

A new interdisciplinary workshop co-located with IJCAI 2024. Analogy-ANGLE will take place on August 4th, 2024 in Jeju, South Korea.

The First Workshop on Analogical Abstraction in Cognition, Perception, and Language (Analogy-ANGLE)

Co-located with

Analogy-ANGLE is a new interdisciplinary workshop co-located with IJCAI 2024 on August 4th, 2024 in Jeju, South Korea. Analogy-ANGLE aims to bring together researchers with an interest in analogical abstraction from natural language processing, cognitive psychology, computer vision, deep learning, and neuro-symbolic AI. This workshop will enable a common ground where the complementary perspectives from these fields can come together to form a comprehensive picture of the current landscape of analogical abstraction, and point to standing challenges, evaluation methodologies, and emerging techniques of interest. Thus, we organize this workshop at IJCAI where leading researchers from the focal areas are gathered. The multidisciplinary nature of the workshop is emphasized by the broad set of skills of the organization team and the program committee, and the diversity principle guiding the list of topics and the invited keynotes.

Confirmed Keynote Speakers

KF
Ken Forbus
Northwestern University
TV
Tony Veale
UC Dublin

Accepted Papers

Novel papers:

Dissemination papers:

Program

09:00 - 09:15 Welcome and walk-in

Session I (Chair: Pia Sommerauer)

09:15 - 09:30 Introduction by Filip Ilievski

09:30 - 10:30 Keynote by Ken Forbus: Analogical Reasoning and Learning for Human-like AI Systems

10:30 - 11:00 Coffee break

Session II (Chair: Pia Sommerauer)

11:00 - 11:20 Donghyeon Shin, Seungpil Lee, Klea Kovačec, Sundong Kim. Regulation Using Large Language Models to Generate Synthetic Data for Evaluating Analogical Ability

11:20 - 11:40 Zhivar Sourati, Filip Ilievski, Pia Sommerauer, Yifan Jiang. ARN: Analogical Reasoning on Narratives

11:40 - 11:50 Philipp Wicke, Lea Hirlimann, João Miguel Cunha. Using Analogical Reasoning to Prompt LLMs for their Intuitions of Abstract Spatial Schemas (video)

11:50 - 12:10 Giulia Rambelli, Emmanuele Chersoni, Claudia Collacciani, Marianna Bolognesi. Can Large Language Models Interpret Noun-Noun Compounds? A Linguistically-Motivated Study on Lexicalized and Novel Compounds

12:10 - 12:20 Howard Goldowsky, Vasanth Sarathy. Analogical Reasoning Within a Conceptual Hyperspace (video)

12:30 - 14:00 Lunch break

Session III (Chair: Filip Ilievski)

14:00 - 15:00 Keynote by Tony Veale: Me, Myself and Irony: Exploring different facets of Irony with Analogies and LLMs

15:00 - 15:20 Giulia Rambelli, Emmanuele Chersoni, Philippe Blache, Alessandro Lenci. Compositionality as an Analogical Process: Introducing ANNE

15:20 - 15:30 Molly Rose Petersen, Lonneke van der Plas. Can Language Models Learn Analogical Reasoning? Investigating Training Objectives and Comparisons to Human Performance (video)

15:30 - 16:00 Coffee break

Session IV

16:00-17:15 Panel/roundtable discussion with the keynote speakers and the paper presenters

17:15-17:30 Closing and final remarks

Workshop Registration

Workshop registration is through the IJCAI conference website - please make sure you select W12 The First Workshop on Analogical Abstraction in Cognition, Perception, and Language (Analogy-ANGLE) when registering.

Topics

We invite contributions ranging from cognitive modeling and algorithms and methods to new tasks and new applications of analogy. Contributions that take an interdisciplinary perspective are particularly encouraged. Topics include (but are not limited to) the following:

Submissions

Submissions can fall into one of the following categories :

  1. Full Research Papers (up to 7 pages plus 2 pages for references) - Papers with original research work which will be judged on their technical soundness and rigor, though allowances made for novel or experimental directions. We also welcome submissions reporting negative results and sharing experimental insights on the technical challenges and issues of analogical abstraction.

  2. Short Papers (up to 4 pages) - Position papers or reports of ongoing work on new research directions.

  3. Dissemination Papers - Already published papers from top AI venues such as IJCAI, NeurIPS, AAAI, ICML, ICLR, ACL, and EMNLP that are relevant to the workshop. Please upload the original submission and abstract to our submission site. Please indicate where the paper has been accepted as a first sentence in the abstract.

Full and short research papers will be peer-reviewed by at least two reviewers from the PC. Accepted full and short papers will be included in the proceedings of the workshop. Dissemination papers will go through a short review from the organizers, checking for their quality and relevance to the workshop. Dissemination papers will not be included in the workshop proceedings.

Please submit your contribution via Chairingtool. The paper type will be inferred based on the submission length. Submissions should be anonymized and the review will be double-blind. Preprints can be stored on arXiv. Analogy-ANGLE has informal proceedings and a flexible policy on double submissions of novel papers.

Please format your full and short papers using the IJCAI template. Dissemination papers can be submitted in their original format.

Selected papers will be invited to submit to the Special Issue on Commonsense Reasoning of the Neurosymbolic Artificial Intelligence journal.

Important Dates

The deadline time is 23:59 anywhere on Earth.

Chairs

FI
Filip Ilievski
VU Amsterdam
Commonsense reasoning,
Natural language processing
PS
Pia Sommerauer
VU Amsterdam
Computational linguistics,
Natural language processing
MB
Marianna Bolognesi
University of Bologna
Cognitive linguistics,
Computational creativity

Contact: analogy-angle@groups.google.com

Advisors

US
Ute Schmid
University of Bamberg
Cognitive science,
Interpretable ML
DS
Dafna Shahaf
The Hebrew University of Jerusalem
Computational creativity,
Data science

Program Committee

Background

Analogical abstraction is a fundamental cognitive skill unique to humans (Penn et al., 2008; Hofstadter, 2001), defined as the ability to perceive and utilize the similarities between concepts, situations or events based on (systems of) relations rather than surface similarities (Holyoak, 2012; Gentner et al., 2012). Analogy enables creative inferences, explanations, and generalization of knowledge, and has been used for scientific inventions (Dunbar and Klahr, 2012), solving problems (Gick and Holyoak, 1980), and policy-making (Houghton, 1998). As such, it has been the goal of one of the first AI programs developed by Evans (1964). It has also been the subject of cognitive theories and studies about humans for common processes, such as the retrieval of memories (Wharton et al., 1994) and problem-solving (Gick and Holyoak, 1980), mostly leveraging narratives as their experimental medium (Gentner and Toupin, 1986; Gentner et al., 1993; Wharton et al., 1994), given their multi-tiered nature and potential for abstraction.

Meanwhile, analogical tasks have also been a relatively popular topic in natural language processing (NLP) and artificial intelligence (AI), typically framed as intelligence tests for models compared against humans. So-called word-based, proportional analogies of the form (A : B :: C : D) (Mikolov et al., 2013a,b; Gladkova et al., 2016; Ushio et al., 2021) are often used to measure the potential of word embeddings and language models. Recent studies (Webb et al., 2023) show a strong ability of state-of-the-art (SOTA) large language models (LLMs) to discover proportional word analogies, though this skill degrades with higher complexity (Wijesiriwardene et al., 2023) or when controlling for association-based answers (Stevenson et al., 2023). Shifting toward more complex settings, narrative-based analogy benchmarks that involve system mappings rather than simple word-based relational mappings have been also been considered recently, with limitations in scope, generalizability, and alignment with cognitive theories (Nagarajah et al., 2022; Wijesiriwardene et al., 2023; Sourati et al., 2023). Meanwhile, given the potential of large language and visual models, another line of research aims to study their ability to draw analogies consistently (cf., Webb et al., 2023). Given the richness of analogical abstraction and the wide interest in this topic from artificial intelligence, linguistics, and cognitive psychology, it is important to connect these communities and facilitate cross-disciplinary activities.

Dunbar, K.N. and Klahr, D., 2012. Scientific thinking and reasoning. In Keith J Holyoak and Robert G Morrison, editors, Oxford handbook of thinking and reasoning, page 701–718.

Evans, T.G., 1964. A program for the solution of a class of geometric-analogy intelligence-test questions (No. 64). Air Force Cambridge Research Laboratories, Office of Aerospace Research, United States Air Force.

Gentner, D., Smith, L. and Ramachandran, V.S., Analogical Reasoning, 2012. Encyclopedia of Human Behavior, 2nd ed., VS Ramachandran, ed., Elsevier, Oxford, UK, pp.130-136.

Gentner, D. and Toupin, C., 1986. Systematicity and surface similarity in the development of analogy. Cognitive science, 10(3), pp.277-300.

Gentner, D., Rattermann, M.J. and Forbus, K.D., 1993. The roles of similarity in transfer: Separating retrievability from inferential soundness. Cognitive psychology, 25(4), pp.524-575.

Gick, M.L. and Holyoak, K.J., 1980. Analogical problem solving. Cognitive psychology, 12(3), pp.306-355.

Gladkova, A. and Drozd, A., 2016, August. Intrinsic evaluations of word embeddings: What can we do better?. In Proceedings of the 1st Workshop on Evaluating Vector-Space Representations for NLP (pp. 36-42).

Hofstadter, D.R., 2001. Analogy as the core of cognition. The analogical mind: Perspectives from cognitive science, pp.499-538.

Holyoak, K.J., 2012. Analogy and relational reasoning. The Oxford handbook of thinking and reasoning, pp.234-259.

Houghton, D.P., 1998. Analogical reasoning and policymaking: Where and when is it used?. Policy Sciences, 31(3), pp.151-176.

Mikolov, T., Chen, K., Corrado, G. and Dean, J., 2013a. Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781.

Mikolov, T., Sutskever, I., Chen, K., Corrado, G.S. and Dean, J., 2013b. Distributed representations of words and phrases and their compositionality. Advances in neural information processing systems, 26.

Nagarajah, T., Ilievski, F. and Pujara, J., 2022. Understanding narratives through dimensions of analogy. IJCAI workshop on Qualitative Reasoning.

Penn, D.C., Holyoak, K.J. and Povinelli, D.J., 2008. Darwin’s mistake: Explaining the discontinuity between human and nonhuman minds. Behavioral and brain sciences, 31(2), pp.109-130.

Stevenson, C.E., ter Veen, M., Choenni, R., van der Maas, H.L. and Shutova, E., 2023. Do large language models solve verbal analogies like children do?. arXiv preprint arXiv:2310.20384.

Sourati, Z., Ilievski, F. and Sommerauer, P., 2024. ARN: A Comprehensive Framework and Dataset for Analogical Reasoning on Narratives. To appear in TACL.

Ushio, A., Espinosa-Anke, L., Schockaert, S. and Camacho-Collados, J., 2022, March. BERT is to NLP what AlexNet is to CV: Can Pre-Trained Language Models Identify Analogies?. In ACL 2022 Workshop on Commonsense Representation and Reasoning.

Webb, T., Holyoak, K.J. and Lu, H., 2023. Emergent analogical reasoning in large language models. Nature Human Behaviour, 7(9), pp.1526-1541.

Wharton, C.M., Holyoak, K.J., Downing, P.E., Lange, T.E., Wickens, T.D. and Melz, E.R., 1994. Below the surface: Analogical similarity and retrieval competition in reminding. Cognitive Psychology, 26(1), pp.64-101.

Wijesiriwardene, T., Sheth, A., Shalin, V.L. and Das, A., 2023. Why Do We Need Neurosymbolic AI to Model Pragmatic Analogies?. IEEE Intelligent Systems, 38(5), pp.12-16.