Selecting a text similarity measure for a content-based recommender system

Abstract

Purpose The purpose of this paper is to develop a journal recommender system, which compares the content similarities between a manuscript and the existing journal articles in two subject corpora (covering the social sciences and medicine). The study examines the appropriateness of three text similarity measures and the impact of numerous aspects of corpus documents on system performance. Design/methodology/approach Implemented three similarity measures one at a time on a journal recommender system with two separate journal corpora. Two distinct samples of test abstracts were classified and evaluated based on the normalized discounted cumulative gain. Findings The BM25 similarity measure outperforms both the cosine and unigram language similarity measures overall. The unigram language measure shows the lowest performance. The performance results are significantly different between each pair of similarity measures, while the BM25 and cosine similarity measures are moderately correlated. The cosine similarity achieves better performance for subjects with higher density of technical vocabulary and shorter corpus documents. Moreover, increasing the number of corpus journals in the domain of social sciences achieved better performance for cosine similarity and BM25. Originality/value This is the first work related to comparing the suitability of a number of string-based similarity measures with distinct corpora for journal recommender systems.

Technological troubleshooting based on sentence embedding with deep transformers

“…By using 100 fine-tuning epochs, we compare the effectiveness of the proposed approach against the one obtained with a differently implemented sentence embedding modules. Specifically, we replace S-BERT with USE (Cer et al, 2018) and employ a well-known bag-of-words approaches, such as the cosine similarity and BM25 (Wijewickrema et al, 2019). USE represents the state of the art for sentence embedding approaches based on transformers (Ahmed et al, 2019).…”

Supplementary and Complementary Congruence at the Apex of the Organization: Management and Board as a Strategic‐Oriented Multiteam System

“…, 2020; Wijewickrema et al. , 2019). Cosine similarity is superior because even if a TMT and board are far apart because of size (boards in our sample are about 50% larger than TMTs), their representing vectors could still have a smaller angle between them, i.e.…”

The Academic Landscapes of Manufacturing Enterprise Performance and Environmental Sustainability: A Study of Commonalities and Differences

“…Cosine similarity with tf-idf is a common and effective strategy to determine the similarity of documents in text mining [ 25 ]. While multiple similarity measures exist, vocabulary comparisons have been found to yield similar results to other metrics, particularly for technical vocabulary and in the social sciences [ 26 ]. Moreover, cosine similarity best approximates the consensus of domain experts when manually comparing cluster content [ 27 ].…”