Reproduce “MobileConvRec: A Conversational Dataset for Mobile Apps Recommendations” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Srijata Maji, Moghis Fereidouni, Vinaik Chhetri, Umar Farooq, A.B. Siddique
The MobileConvRec dataset addresses the absence of high-quality conversational datasets for mobile app recommendations by providing over 12,200 multi-turn dialogues derived from historical app usage and simulated conversations. This dataset enables the development of app recommender systems that consider both implicit user preferences and explicit conversational input, with baseline experiments demonstrating its utility for app name and response generation tasks.
Key insights
- The dataset integrates both implicit preferences (historical app usage) and explicit needs (conversational input) for more nuanced recommendations.
- Conversations generated using GPT-3.5 Turbo are multi-turn (typically 12-14 turns) and feature natural language patterns with rich vocabulary.
- The inclusion of app metadata like permissions and executable files allows for research into security and privacy aspects of mobile apps.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.