Artificial Intelligence, Wikipedia and Semi-Structured Resources

The availability of large amounts of wide-coverage semantic knowledge, and the ability to extract it using the powerful new statistical machine learning techniques developed and used in various branches of AI, is making possible significant advances in applications that require deep understanding capabilities such as question-answering engines and dialogue systems. Though well-known problems such as high cost and scalability discouraged the development of knowledge-rich approaches in the past, more recently the increasing availability of online collaborative resources has attracted the attention of much work in the AI community. Collaboratively constructed knowledge repositories have in fact been used as wide-coverage sources of semi-structured information and manual annotations. When coupled with free-form natural language information, these resources enable the development of large-scale structured resources using knowledge-lean applications. Wikipedia is a case in point, being the largest and most popular collaborative and multilingual resource of world and linguistic knowledge that contains unstructured and (semi-) structured information.

This special issue aims to collect state-of-the-art contributions to the development and use of hybrid (structured, semi-structured, and unstructured) resources in AI. These include, but are not limited to, semi-structured encyclopedic resources such as Wikipedia (and related projects such as Wiktionary), user-generated answer repositories such as Wiki and Yahoo! Answers, and collaborative tagging efforts on social media platforms such as Flickr and Blogger. Hybrid knowledge resources such as Wikipedia enable the development of methods for extracting, bootstrapping and integrating fully structured, machine-readable knowledge from both unstructured and semi-structured origins. Such induced wide-coverage knowledge is expected to prove beneficial for a variety of AI tasks, as well as the Semantic Web. We are particularly interested in articles showing the benefits of using such resources and AI techniques synergistically. We thus welcome contributions dealing with applications of general AI methodologies for the construction and validation of large-scale machine-readable knowledge repositories and the impact of automatically-extracted knowledge for AI applications. We also encourage contributors to investigate the nature and impact of the structured and unstructured parts of the resource (e.g. information redundancy, overlaps, connections, etc.).