Location Aware Modular Biencoder for Tourism Question Answering
This addresses the problem of scalable Point-of-Interest recommendation for tourism applications, though it is incremental as it adapts existing dense retrieval techniques to this domain.
The paper tackled the challenge of efficiently answering tourism questions requiring spatial and non-spatial reasoning by proposing a dense vector retrieval approach, which outperformed previous methods across all metrics and expanded the search space by 20 times.
Answering real-world tourism questions that seek Point-of-Interest (POI) recommendations is challenging, as it requires both spatial and non-spatial reasoning, over a large candidate pool. The traditional method of encoding each pair of question and POI becomes inefficient when the number of candidates increases, making it infeasible for real-world applications. To overcome this, we propose treating the QA task as a dense vector retrieval problem, where we encode questions and POIs separately and retrieve the most relevant POIs for a question by utilizing embedding space similarity. We use pretrained language models (PLMs) to encode textual information, and train a location encoder to capture spatial information of POIs. Experiments on a real-world tourism QA dataset demonstrate that our approach is effective, efficient, and outperforms previous methods across all metrics. Enabled by the dense retrieval architecture, we further build a global evaluation baseline, expanding the search space by 20 times compared to previous work. We also explore several factors that impact on the model's performance through follow-up experiments. Our code and model are publicly available at https://github.com/haonan-li/LAMB.