With 102,530,067 items currently in its crowd-sourced knowledge base, Wikidata provides NLP practitioners a unique and powerful resource for inference and reasoning over real-world entities. However, because Wikidata is very entity focused, events and actions are often labeled with eventive nouns (e.g., the process of diagnosing a person’s illness is labeled “diagnosis”), and the typical participants in an event are not described or linked to that event concept (e.g., the medical professional or patient). Motivated by a need for an adaptable, comprehensive, domain-flexible ontology for information extraction, including identifying the roles entities are playing in an event, we present a curated subset of Wikidata in which events have been enriched with PropBank roles. To enable richer narrative understanding between events from Wikidata concepts, we have also provided a comprehensive mapping from temporal Qnodes and Pnodes to the Allen Interval Temporal Logic relations.

Natural language analysis of patents holds promise for the development of tools designed to assist analysts in the monitoring of emerging technologies. One component of such tools is the identification of technology terms. We describe an approach to the discovery of technology terms using supervised machine learning and evaluate its performance on subsets of patents in three languages: English, German, and Chinese.

We present an approach to the discovery of semantically similar terms that utilizes a web search engine as both a source for generating related terms and a tool for estimating the semantic similarity of terms. The system works by associating with each document in the search engine’s index a weighted term vector comprising those phrases that best describe the document’s subject matter. Related terms for a given seed phrase are generated by running the seed as a search query and mining the result vector produced by averaging the weights of terms associated with the top documents of the query result set. The degree of similarity between the seed term and each related term is then computed as the cosine of the angle between their respective result vectors. We test the effectiveness of this approach for building a term recommender system designed to help online advertisers discover additional phrases to describe their product offering. A comparison of its output with that of several alternative methods finds it to be competitive with the best known alternative.