Multi-lingual neural title generation for e-commerce browse pages

Abstract

To provide better access of the inventory to buyers and better search engine optimization, e-Commerce websites are automatically generating millions of easily searchable browse pages. A browse page groups multiple items with shared characteristics together. It consists of a set of slot name/value pairs within a given category that are linked among each other and can be organized in a hierarchy. This structure allows users to navigate laterally between different browse pages (i.e. browse between related items) or to dive deeper and refine their search. These browse pages require a title describing the content of the page. Since the number of browse pages is huge, manual creation of these titles is infeasible. Previous statistical and neural generation approaches depend heavily on the availability of large amounts of data in a language. In this research, we apply sequence-tosequence models to generate titles for high- & low-resourced languages by leveraging transfer learning. We train these models on multilingual data, thereby creating one joint model which can generate titles in various different languages. Performance of the title generation system is evaluated on three different languages; English, German, and French, with a particular focus on low-resourced French language.