An effective, low-cost measure of semantic relatedness obtained from Wikipedia links

By David N. Milne, Ian H. Witten. 2008

809 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/3e48cb9b4838f8947388c3926df70211ca49b51a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Local and Global Algorithms for Disambiguation to Wikipedia (2011, methodology): “The first data set, from (Milne and Witten, 2008b), is a subset of the AQUAINTcorpus of newswire text that is annotated to mimic the hyperlink structure in Wikipedia.”
Methods for exploring and mining tables on Wikipedia (2013, methodology): “As one example, the semantic relatedness (SR) between Wikipedia pages estimated from the link graph [11] forms an extremely valuable feature for identifying relevant joins.”
A framework for benchmarking entity-annotation systems (2013, background): “Disambiguation exploits the structure of the Wikipedia graph, according to the relatedness measure introduced in [17] which takes into account the amount of common incoming links between two pages.”
Fine-Grained Entity Typing for Domain Independent Entity Linking (2019, methodology): “Preneural systems used link structure (Milne and Witten, 2008) and tf-idf (Ratinov et al.”
KORE: keyphrase overlap relatedness for entity disambiguation (2012, methodology): “Most notably, the Milne-Witten measure considers the overlap of the incoming links to two entity articles as a notion of their relatedness [22].”
Integrating Cyc and Wikipedia: Folksonomy meets rigorously defined common-sense (2008, background): “of candidate Wikipedia articles and a set of related context articles, we determine the candidate that is most semantically related to a given context ( Milne and Witten, 2008 ).”
Leveraging Deep Neural Networks and Knowledge Graphs for Entity Disambiguation (2015, methodology): “(ii) Is the DNN a better choice than Normalized Google Distance (NGD) (Cilibrasi and Vitanyi, 2007) and Vector Space Model (VSP) (Salton et al., 1975) for relatedness measurement?”
Polarity Inducing Latent Semantic Analysis (2012, background): “Structured information can be important in determining relatedness, and thesauri and Wikipedia links have been studied in (Milne and Witten, 2008; Jarmasz and Szpakow-icz, 2003).”
Robust Entity Linking via Random Walks (2014, methodology): “Another approach uses the set of entities directly connected in the KB [20].”
A Survey on Context Learning (2017, background): “Some works like [20], [21], [22], [23], [24], [25] makes use of exter-”
Joint Learning of the Embedding of Words and Entities for Named Entity Disambiguation (2016, methodology): “Huang et al. (Huang et al., 2015) also leveraged deep neural networks to learn entity representations such that the consequent pairwise entity relatedness was more suitable than of a standard method (i.e., WLM) for NED.”
A New Set of Norms for Semantic Relatedness Measures (2013, methodology): “Several studies have reported correlation to WordSim353 norms as part of their evaluation procedures, with some studies explicitly referring to it as a collection of human-assigned relatedness scores (Gabrilovich and Markovitch, 2007; Hughes and Ramage, 2007; Milne and Witten, 2008).”
Automatic Evaluation of Topic Coherence (2010, methodology): “In the last few years, there has been a surge of interest in usingWikipedia to calculate semantic similarity, using theWikipedia article content, in-article links and document categories (Str übe and Ponzetto, 2006; Gabrilovich and Markovitch, 2007; Milne and Witten, 2008).”