An information-theoretic perspective of tf-idf measures

By Akiko Aizawa. Information Processing & Management, 2003

1,384 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems

View Paper: https://www.semanticscholar.org/paper/9b9a7642c9018deb8d66455f14ed38930b7d2e59

How researchers see this paper

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

In Artificial Intelligence

In Information Systems

Citations in the corpus

Understanding inverse document frequency: on theoretical arguments for IDF (2004, background): “All these are closely related (see e.g. Aizawa, 2003).”
Modeling word burstiness using the Dirichlet distribution (2005, background): “One traditional information retrieval heuristic is the termfrequency inverse-document-frequency (TFIDF) transformation, which exists in various forms (Aizawa, 2003).”
Interpreting TF-IDF term weights as making relevance decisions (2008, background): “IDF is still a subject of current research [Joachims 1997; Amati and van Rijsbergen 1998; Hiemstra 1998; Papineni 2001; Aizawa 2003; Roelleke 2003; Lee 2007] where Robertson [2004] and Sp¨ arck Jones [2004] responded to recent developments on interpreting IDF.”
N-gram IDF: A Global Term Weighting Scheme Based on Information Distance (2015, methodology): “In addition to the effectiveness of TF-IDF in various applications such as information retrieval [46], document clustering [14], key term extraction [19] and object matching in videos [44], many theoretical explanations [2, 20, 36, 38] encourage researchers and developers to employ TF-IDF.”
When does deep multi-task learning work for loosely related document classification tasks? (2018, background): “This representation is known to be efﬁcient (Salton and Buckley, 1988; Aizawa, 2003); especially in the case of text classiﬁcation (Zhang, Yoshida, and Tang, 2011).”
Sensing spatial distribution of urban land use by integrating points-of-interest and Google Word2Vec model (2017, background): “By comparing with some state-of-the-art topic models (TF-IDF, pLSA and LDA), the proposed method was applied to the classi ﬁ cation of urban land use and obtained the highest accuracy and e ﬃ ciency (OA = 0.8728, kappa = 0.8399, average computation time = 161.0040 s ).”
Application of information retrieval techniques to single writer documents (2005, methodology): “The tf Æ idf is the most applied weighting scheme (Baeza-Yates and Ribeiro-Neto, 1999; Aizawa, 2003) and it embodies the intuition that the more a term appears in a document, the more it is representative of its content and that terms appearing in few documents allow better discrimination between di”