Hybrid approach improves finding similar mobile apps with language models
ReFilter: Bridging Embeddings and LLM Filtering for Similar Mobile App Retrieval
Information Retrieval
Summary
Finding apps that do the same thing can be hard because apps that look different can actually work similarly. The authors created a new system called ReFilter that first uses a quick way to find apps that seem related and then uses a smarter language model to check which ones really work the same. This makes finding similar apps more accurate without needing too much computer power. It helps people compare apps better and get smarter recommendations.
What this means in practice
- •For mobile app developers: Identify apps performing similar functions to analyze competitors more accurately using a combination of embedding retrieval and language model filtering.
- •For ecommerce recommendation teams: Enhance app recommendation systems by filtering semantically related apps to find truly functionally similar alternatives for end-users.$Commercial implications: Enables more precise app recommendations for app stores and marketplaces, improving user satisfaction and sales.
Authors
Buthayna AlMulla, Maram Assi, Safwat Hassan
Abstract
Retrieving similar mobile applications (apps) is essential for researchers, developers, and end-users. Researchers use similarity detection to study app ecosystems and trends, developers for competitor analysis, and end-users for focused app recommendations. Existing approaches rely on embedding-based retrieval, which captures semantic similarity but fails to identify functionally similar apps. To our knowledge, no prior work has applied large language model (LLM)-based filtering to this task, due to the high computational cost of evaluating large numbers of app pairs. To address this gap, we propose ReFilter, a hybrid framework that first Retrieves semantically related candidate apps using embeddings and then applies LLM-based contextual Filtering to identify true functionally similar apps with higher precision. This design balances efficiency and accuracy, achieving an F1-score of 90% for retrieving similar apps. By improving the relevance of app alternatives, ReFilter enables more accurate app comparisons and supports improved ecosystem understanding, competitor analysis, and recommendations.