Wikification via link co-occurrence

Abstract

Wikification, which stands for the process of linking terms in a plain text document to Wikipedia articles which represent the correct meanings of the terms, can be thought of as a generalized Word Sense Disambiguation problem. It disambiguates multi-word expressions (MWEs) in addition to single words. Existing Wikification techniques either models the context of a given term as well as the Wikipedia article as bags of words, or compute global constraints among Wikipedia concepts by the link graph or link distributions. The first method doesn't achieve good results because the MWEs can have very different meanings than its constituent words which themselves are ambiguous. The second method doesn't produce high accuracy because the link structure or link distribution is often biased or incomplete by themselves due to the fact that Wikipedia pages are often sparsely linked. In this paper, we present a simple but powerful framework of sense disambiguation using cooccurrences of Wikipedia links in the Wikipedia corpus. We propose an iterative method to enrich the sparsely-linked articles by adding more links and then use the resulting link co-occurrence matrix to disambiguate an input document by a sliding window algorithm. Our prototype system achieves 89.97% precision and 76.43% recall on average for three benchmark data sets and compares favorably against all four state-of-the-art wikification techniques.1

Exploring graph topology via matrix factorization to improve wikification

“…By doing so, we hope to improve on previous feature-and topology-based methods. To the our best knowledge, only Cai et al [4] and West wt al. [15] directly used topological aspects of the Wikipedia concept graph for wikification.…”

“…It is important to observe that, in this approach, wikification is treated as a single task and disambiguation is performed implicitly. In this section, we summarize unsupervised strategies that are representatives of this approach [2,4,15].…”

“…This way, an error reconstruction per link can be assessed and used to rank potential new links. Cai et al [4] developed an iterative algorithm that enriches sparsely-linked articles by adding more links. Their algorithm makes use of Wikipedia links co-occurrence matrix to provide these links at each iteration.…”

Dawt

“…To facilitate studies using Wikidata ids, which are also widely used entity ids in the literatures, we provide a data set that maps individual Freebase ids to Wikidata ids. This data set contains twice as many mappings as that from Google 3 . A summary comparison between these two mapping sets are shown in Table 6, which lists the total numbers of mappings in 4 buckets:…”

“…The Freebase annotations of the ClueWeb corpora dataset 8 [5,7,19] while in global approach entities across the entire document are disambiguated together using document context, thus, ensuring consistency of entities across the document [9,18]. Most recently Cai et al [3] achieved 89.97% precision and 76.43% recall, using an iterative algorithm that leverages link graph, link distributions, and a noun phrase extractor.…”