Knowledge Base Construction

Knowledge gathering, representation, and reasoning are among the fundamental challenges of artificial intelligence. Large-scale repositories of knowledge about entities, relations, and their abstractions are known as “knowledge bases”. Most major technology companies now have substantial efforts in knowledge base construction. Related scholarly work spans many research areas, including machine learning, natural language processing, computer vision, information integration, databases, search, data mining, knowledge representation, human computation, human-computer interfaces, and fairness. The AKBC conference serves as a research forum for all these areas, in both academia and industry.
New Conference

Nearly ten years after the first AKBC workshop in Grenoble, France, AKBC has become a conference, with its widely-acclaimed first instantiation last year in Amherst MA. Why a new stand-alone conference?

    Long-standing and growing interest in the area, now with too much material for a one-day workshop. We have sufficient material for a two-day conference plus topical workshops.
    We want to grow and connect the community beyond existing individual conference communities, bringing together ML, NLP, DB, IR, KRR, semantics, reasoning, common sense, QA, human computation, dialog, HCI.
    We want to set our own culture, including reviewing practices, and meeting format. We have fond memories of the first AKBC 2010 in Grenoble: a two-day meeting that included an afternoon hike in the Alps with much great scientific discussion.
    Why now? Growing interest across many areas. Disconnect among multiple relevant communities. Growing industry and government interest. Many of the long-existing conferences have grown uncomfortably large; a new, smaller conference can be more intimate, hospitable, and supportive.

Call for Papers

We invite the submission of papers describing previously unpublished research, including new methodology, datasets, evaluations, surveys, reproduced results, negative results, and visionary positions.

Topics of interest include, but are not limited to:

    Natural language processing, information extraction, extraction of entities, relations, and events, semantic parsing, coreference, machine reading, entailment, web mining, multilingual NLP.
    Information integration, entity resolution, schema & ontology alignment, text and structure alignment, federated KBs, Semantic Web.
    Machine learning, supervised, unsupervised, lightly-supervised and distantly-supervised learning, deep learning, symbolic learning, multimodal learning, embeddings of knowledge.
    Search, question-answering, reasoning, knowledge base completion, queries on mixtures of structured and unstructured data; querying under uncertainty.
    Multi-modal knowledge bases: structured data, text, images, video, audio.
    Human-computer interaction, crowdsourcing, interactive learning.
    Fairness, accountability, transparency, misinformation, multiple viewpoints, uncertainty.
    Databases, probabilistic databases, distributed databases, database cleaning, scalable computation, distributed computation, dynamic data, online adaptation of knowledge.
    Systems, languages and toolkits, demonstrations of existing knowledge bases.
    Evaluation of AKBC, datasets, evaluation methodology.

Authors of accepted papers will have the option for their conference paper to be archival (with full text in AKBC Proceedings, and be considered for best paper awards) or non-archival (listed in AKBC Conference schedule, with full text in OpenReview, and the flexibility to also submit elsewhere). Double-blind reviewing will be performed on the OpenReview platform, with papers, reviews and comments publicly visible, much like ICLR 2020.

Papers should be restricted to 10 pages excluding references (the equivalent of about 8 pages double column).