Mobile app development necessitates extracting domain-specific, essential, and innovative features that align with user needs and market trends. Determining which features provide a competitive advantage is a complex task, often managed manually by product managers. This study addresses the challenge of automating feature mining and recommendation by identifying similar apps based on user-provided descriptions. The proposed approach integrates Named Entity Recognition (NER) for feature extraction from mined Google Play app data with BERT (Bidirectional Encoder Representations from Transformers) and Topic Modeling to find comparable apps. Our top-performing model, which uses Non-negative Matrix Factorization (NMF) for Topic Modeling with Sentence-BERT (SBERT) embeddings, achieves an F1 score of 87.38%.

Context-aware recommender system (CARS) plays a vital role to paved way for improving traditional recommendation phenomena. A key issue in CARS is the selection of influencing contextual group which is most suitable for the recommendation of an item. In general, the selection of suitable contextual group in CARS is faced with challenges due to uncertainty in the classification of items with missing non-binary ratings. The underlying reaction of each user towards an item is implicitly assumed to be binary in nature because of which uncertainty occurs in the item classification. In particular, such a situation where ratings are missing for an item, make the exploitation of appropriate contextual group difficult for an item recommendation. In this article, we address the problem of the inappropriate classification of items into irrelevant contextual groups due to missing non-binary ratings. To this extent, we propose a three-way classification model using game-theoretic N-soft sets for improving the classification process by handling missing ratings. In particular, a game is formulated using game-theoretic N-soft sets to determine the effective threshold configuration used to induce three-way classification of items with missing non-binary ratings. Moreover, a thorough evaluation of our proposed model is carried out on the datasets of LDOS-CoMoDa and InCarMusic, where outcome signifies the effectiveness and performance of the proposed model.

The adoption of the contextual information in recommender systems is fairly recent. Context-Aware Recommender Systems (CARS) provide better-personalized recommendations by utilizing contextual features in comparison to the classical two-dimensional recommendation process. In addition, selection of the specific recommender algorithm has a direct impact on the performance of CARS. In the context of video recommender systems, roughness exist in deciding which context is better to choose when a user had watched same video in more than one contextual scenarios. Therefore, selection of appropriate context and recommender algorithm for developing CARS is identified as potential research challenge. This problem may be investigated by considering a formal approximation on contextual attributes. In this article, we investigate the applicability of using soft-rough sets for formulating such an approximation on a sample real world scenario. The main objective of this study is demonstrating the applicability of soft-rough sets on CARS for removing roughness and improving contextual information selection process. The experimental results revealed that only 27% of the rules were identified by applying Rough sets on context-aware video recommender systems (CAVRS). These results can be further improved using Soft-rough sets by representing the given scenario in Boolean-valued information systems. Additionally, a few recommendations for future work on CARS are proposed.

Software fault localisation (SFL) is recognised to be one of the most tedious, costly, and critical activities in program debugging. Due to the increase in software complexity, there is a huge interest in advanced SFL techniques that aid software engineers in locating program bugs. This interest paves a way to the existence of a large amount of literature in the SFL research domain. This study aims to investigate the overall research productivity, demographics, and trends shaping the landscape of SFL research domain. The research also aims to classify existing fault localisation techniques and identify trends in the field of study. Accordingly, a systematic mapping study of 273 primary selected studies is conducted with the adoption of an evidence‐based systematic methodology to ensure coverage of all relevant studies. The results of this systematic mapping study show that SFL research domain is gaining more attention since 2010, with an increasing number of publications per year. Three main research facets were identified, i.e. validation research, evaluation research, and solution research, with solution research type getting more attention. Hence, various contribution facets were identified as well. In totality, general demographics of SFL research domain were highlighted and discussed.