Empirical comparison of text-based mobile apps similarity measurement techniques

Abstract

Context Code-free software similarity detection techniques have been used to support different software engineering tasks, including clustering mobile applications (apps). The way of measuring similarity may affect both the efficiency and quality of clustering solutions. However, there has been no previous comparative study of feature extraction methods used to guide mobile app clustering. Objective In this paper, we investigate different techniques to compute the similarity of apps based on their textual descriptions and evaluate their effectiveness using hierarchical agglomerative clustering. Method To this end we carry out an empirical study comparing five different techniques, based on topic modelling and keyword feature extraction, to cluster 12,664 apps randomly sampled from the Google Play App Store. The comparison is based on three main criteria: silhouette width measure, human judgement and execution time. Results The results of our study show that using topic modelling, in addition to collocationbased and dependency-based feature extractors perform similarly in detecting app-feature similarity. However, dependency-based feature extraction performs better than any other in finding application domain similarity (ρ = 0.7, p − value < 0.01).

Exploring the Use of Genetic Algorithm Clustering for Mobile App Categorisation

“…The original study extracted features from mobile apps descriptions using four different techniques and compared their effectiveness for clustering apps by using these features and only one clustering approach, i.e. hierarchical clustering [1]. The study found that extracting features using Latent Dirichlet Allocation (LDA) consistently performs well among the investigated feature extraction techniques.…”

“…In order to answer these RQs, we used the same dataset as the original study [1] 3 . This dataset contains 12,664 Android mobile applications belonging to 24 categories, which have been randomly sampled from the Google Play app store.…”

“…This dataset contains 12,664 Android mobile applications belonging to 24 categories, which have been randomly sampled from the Google Play app store. A detailed description of how this data was collected can be found elsewhere [1]. In this study, we opted to use the GA clustering approach proposed by Maulik and Bandyopadhyay [13].…”

“…In this paper, we carry out a partial replication of the original study [1] to investigate whether the results can be improved using a Genetic Algorithm-based clustering algorithm, as evolutionary approaches were shown to be successful as clustering techniques in other application domains [10]. Specifically, we investigate four of the five research questions posed in the original study, but we shift the focus on the clustering approach rather than the feature extraction method.…”

A Text Similarity Measurement Based on Semantic Fingerprint of Characteristic Phrases

“…In terms of large-scale similarity detection based on slightly modified digital fingerprints for Chinese, there is currently no similar research at home and abroad. A. Al-Subaihin [20] proposed a new approach to batch text similarity detection is proposed by combining some ideas from dimensionality reduction techniques and information gain theory. It was focused on search engines need to detect duplicated and near-duplicated web pages.…”