Using Information Content to Evaluate Semantic Similarity in a Taxonomy

By Philip Resnik. 1995

This paper presents a new measure of semantic similarity in an IS-A taxonomy, based on the notion of information content. Experimental evaluation suggests that the measure performs encouragingly well (a correlation of r = 0.79 with a benchmark set of human similarity judgments, with an upper bound of r = 0.90 for human subjects performing the same task), and significantly better than the traditional edge counting approach (r = 0.66).

Subfields: Information Retrieval & Search

How researchers see this paper

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

In Information Retrieval & Search

Citations in the corpus

From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions (2014, methodology): “It consists of a log-linear regression model trained on multiple text features (word and character n-grams, longest common substring and longest common subsequence, Gabrilovich and Markovitch (2007)’s Explicit Semantic Analysis, and Resnik (1995)’s WordNet-based similarity).”
From Frequency to Meaning: Vector Space Models of Semantics (2010, methodology): “In computational linguistics, the term semantic similarity is applied to words that share a hypernym (car and bicycle are semantically similar, because they share the hypernym vehicle) (Resnik, 1995).”
SimLex-999: Evaluating Semantic Models With (Genuine) Similarity Estimation (2014, methodology): “This is supported by the fact that the WordNet-based methods that perform best at modeling human ratings model the similarity between concepts c1 and c2 as the minimum of all pairwise distances between the senses of c1 and the senses of c2 (Resnik, 1995; Pedersen et al., 2004).”
GOATOOLS: A Python library for Gene Ontology analyses (2018, methodology): “The DAG data structure allows traversal of terms along the hierarchy for tasks such as determination of level or depth, retrieval of parent or child terms, and calculation of semantic similarities (e.g. Resnik’s score 13 and Lin’s score 14 ) between terms.”
SemEval-2016 Task 1: Semantic Textual Similarity, Monolingual and Cross-Lingual Evaluation (2016, methodology): “Specifically, MayoNLPTeam combines a pipeline that makes use of linguistic resources such as WordNet and well understood concepts such as the information content of a word (Resnik, 1995) with a deep learning method known as Deep Structured Semantic Model (DSSM) (Huang et al.”
Evaluation of output embeddings for fine-grained image classification (2014, methodology): “In this case, the distance between nodes is measured using semantic similarity metrics [24, 30, 32, 44].”
Domain and Function: A Dual-Space Model of Semantic Relations and Compositions (2012, background): “The results in Section 4.4 suggest that function similarity may correspond to the kind of taxonomical similarity that is often associated with lexicons, such as WordNet (Resnik, 1995; Jiang & Conrath, 1997; Leacock & Chodrow, 1998; Hirst & St-Onge, 1998).”
Robust Textual Inference via Graph Matching (2005, methodology): “In particular, we use the measure described in (Resnik, 1995).”
Context-theoretic Semantics for Natural Language: an Overview (2009, background): “This is an idea that has come about through the introduction of “distance measures” on taxonomies (Resnik, 1995).”
Dependency-Based Construction of Semantic Space Models (2007, methodology): “…(1965) is routinely used in NLP and cognitive science for development purposes—for example, for evaluating automatic measures of semantic similarity (Resnik 1995; Budanitsky and Hirst 2001; Banerjee and Pedersen 2003) or for exploring the parameter space of vector space models (McDonald 2000).”
Multiple SVM-RFE for gene selection in cancer classification with expression data (2005, background): “Several information content-based measures have been proposed [18]–[20], which can be applied to GO-based similarity estimation.”
Query Expansion Techniques for Information Retrieval: a Survey (2017, methodology): “d(s1) + d(s2) Resnik 1995 [216] Resnik similarity −log p(LCS(s1, s2))”
The Human Phenotype Ontology: a tool for annotating and analyzing human hereditary disease. (2008, background): “Resnik 18 introduced a similarity measure for two terms in an ontology that is based on their shared information content, which is given by the information content in the set of their common-ancestor nodes.”
A Topic Model for Word Sense Disambiguation (2007, methodology): “Our approach gives a probabilistic method of using information content (Resnik, 1995) as a starting point that can be adjusted to cluster words in a given topic together; this is similar to the JiangConrath similarity measure (1997), which has been used in many applications in addition to disambigua”
Automatically Creating Datasets for Measures of Semantic Relatedness (2006, methodology): “This experiment was again replicated by Resnik (1995) with 10 subjects.”
L2F/INESC-ID at SemEval-2017 Tasks 1 and 2: Lexical and semantic features in word and textual similarity (2017, background): “…, where: • syn ( t ) gives the synset of the token t ; • hyp ( t ) gives the hypernyms of t ; , with n and m being the number of nodes traversed in the synsets of t 1 and t 2 , respectively. equiv WN matches, thus, two tokens if they have a common hypernym (Resnik, 1995) in their synset path.”
From Word to Sense Embeddings: A Survey on Vector Representations of Meaning (2018, methodology): “Ettinger, Resnik, and Carpuat (2016) followed the same sense induction procedure but used the retrofittingbased sense representation of Jauhar, Dyer, and Hovy (2015) 11, by replacing the standard sense inventory used in the original model (WordNet) with a translation-specific inventory.”