FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W118771212> ?p ?o ?g. }

• W118771212 abstract "Events in the world around us are frequently discussed in natural language, and their accurate identification is central to tasks from intelligence analytics to question answering. They are nonetheless very diverse, both in ontology and in how they are referred to, and have complex but ill-defined structures. For these reasons, computationally identifying and characterising events by how they are referred to proves very challenging. We establish this argument through a broad survey of computational tasks that identify and characterise references to events. News and social media play an important role in informing the public of events as they occur, thus providing a shared foundation for communication. This work presents several studies into interpreting references to the breadth of events covered in news, before proposing and evaluating a new model for computationally grounding newsworthy event references. We perform two annotation studies over broad-coverage news text, the first applying a coarse-grained event typology to mark event-referring sentences. Its results concur with our analysis of inter-annotator agreement and type distribution in the ace05 corpus (Walker et al., 2006): both highlight the brittleness of such type schemas which leave a long, heavy tail of news events unmarked. The second annotation introduces a new approach to event typology, which employs a hierarchy of types that is extended over the course of annotation. This yields more complete coverage of newsworthy events, but suggests that there is no definitive means of structuring the space of events. The second annotation is also novel in characterising each news report in terms of its update event and topic event, focusing only on the notable event content of news. We find that topic is too ambiguously characterised as a single event, suggesting that we should instead consider the explicit references to related events that a journalist provides as background. Despite Wikipedia’s utility for processing entity references, an analysis of its event articles shows that they are unsuitable for informing news event detection and disambiguation: they are strongly biased towards those that are enumerable (e.g. sports series), or that are collections of newsworthy sub-events (e.g. Vietnam War); their distribution, granularity and referential forms are mismatched to news media’s event coverage. In this context, we propose the event linking task. By analogy with named entity linking or disambiguation (Bunescu and Pasca, 2006), event linking models the grounding of references to notable events. It treats a news archive as a proxy for the set of events it reports, and defines the disambiguation of a newsworthy event reference as a link to the article that first reports it. When two references are linked to the same article, they need not be references to the same event. We argue that precise event coreference is often too strict: it does not account for the intricate structure of events, nor the flexibility of referential language. By considering events at the granularity in which they are reported, we hope to provide a more intuitive" @default.
• W118771212 created "2016-06-24" @default.
• W118771212 creator A5079518059 @default.
• W118771212 date "2013-08-31" @default.
• W118771212 modified "2023-09-23" @default.
• W118771212 title "Grounding event references in news" @default.
• W118771212 cites W109605897 @default.
• W118771212 cites W111277439 @default.
• W118771212 cites W11298561 @default.
• W118771212 cites W117509291 @default.
• W118771212 cites W1440493617 @default.
• W118771212 cites W144269646 @default.
• W118771212 cites W146925317 @default.
• W118771212 cites W1493490255 @default.
• W118771212 cites W1493726997 @default.
• W118771212 cites W1500614872 @default.
• W118771212 cites W1502876877 @default.
• W118771212 cites W151066029 @default.
• W118771212 cites W1519551889 @default.
• W118771212 cites W1528445643 @default.
• W118771212 cites W1529628505 @default.
• W118771212 cites W1532325895 @default.
• W118771212 cites W1548663377 @default.
• W118771212 cites W1550258693 @default.
• W118771212 cites W1555467355 @default.
• W118771212 cites W1570278093 @default.
• W118771212 cites W1572567744 @default.
• W118771212 cites W157725869 @default.
• W118771212 cites W157776793 @default.
• W118771212 cites W1594112393 @default.
• W118771212 cites W1596410906 @default.
• W118771212 cites W1601617543 @default.
• W118771212 cites W160393635 @default.
• W118771212 cites W1604485367 @default.
• W118771212 cites W1612381393 @default.
• W118771212 cites W1620608722 @default.
• W118771212 cites W1741471588 @default.
• W118771212 cites W1755857849 @default.
• W118771212 cites W1787402258 @default.
• W118771212 cites W1811279891 @default.
• W118771212 cites W1823704095 @default.
• W118771212 cites W189596627 @default.
• W118771212 cites W1921650316 @default.
• W118771212 cites W1964719994 @default.
• W118771212 cites W1970381522 @default.
• W118771212 cites W1970849810 @default.
• W118771212 cites W1971563386 @default.
• W118771212 cites W1975673151 @default.
• W118771212 cites W1976899842 @default.
• W118771212 cites W1978394996 @default.
• W118771212 cites W1978672522 @default.
• W118771212 cites W1981500266 @default.
• W118771212 cites W1987244517 @default.
• W118771212 cites W1989785251 @default.
• W118771212 cites W1991018417 @default.
• W118771212 cites W1992665562 @default.
• W118771212 cites W1997773542 @default.
• W118771212 cites W1998224037 @default.
• W118771212 cites W1999315683 @default.
• W118771212 cites W2000363133 @default.
• W118771212 cites W2000900121 @default.
• W118771212 cites W2001856540 @default.
• W118771212 cites W2004384146 @default.
• W118771212 cites W2004803990 @default.
• W118771212 cites W2005492507 @default.
• W118771212 cites W2005580324 @default.
• W118771212 cites W2006941876 @default.
• W118771212 cites W2007291639 @default.
• W118771212 cites W2007760849 @default.
• W118771212 cites W2008011355 @default.
• W118771212 cites W2008830554 @default.
• W118771212 cites W2010581447 @default.
• W118771212 cites W2012179495 @default.
• W118771212 cites W2015707701 @default.
• W118771212 cites W2016522586 @default.
• W118771212 cites W2017875634 @default.
• W118771212 cites W2021836477 @default.
• W118771212 cites W2026302857 @default.
• W118771212 cites W2029915058 @default.
• W118771212 cites W2032566933 @default.
• W118771212 cites W2049347837 @default.
• W118771212 cites W2052408961 @default.
• W118771212 cites W2055294489 @default.
• W118771212 cites W2056664748 @default.
• W118771212 cites W2058070641 @default.
• W118771212 cites W2068882115 @default.
• W118771212 cites W2072628044 @default.
• W118771212 cites W2076008912 @default.
• W118771212 cites W2076578255 @default.
• W118771212 cites W2080527295 @default.
• W118771212 cites W2083055089 @default.
• W118771212 cites W2086925418 @default.
• W118771212 cites W2089508313 @default.
• W118771212 cites W209114341 @default.
• W118771212 cites W2091181883 @default.
• W118771212 cites W2094728533 @default.
• W118771212 cites W2096335387 @default.
• W118771212 cites W2097123411 @default.
• W118771212 cites W2101268022 @default.
• W118771212 cites W2101816065 @default.

• next