TestMySearch Blog

This paper addresses a significant challenge in e-commerce information retrieval: the failure of standard keyword search systems to correctly interpret complex user queries that contain product attributes. Queries such as “blue XL Burton jacket” are often processed as a simple set of keywords, leading to irrelevant results and compelling users to engage in a laborious manual filtering process. We present a proof-of-concept (PoC) for an automatic facet discovery system designed to parse user queries, identify terms corresponding to product facets (e.g., color, brand, size), and apply these filters automatically. This research demonstrates a practical methodology for bridging the semantic gap between unstructured free-text search and structured faceted navigation, thereby enhancing result relevance and improving the overall user experience.

Today I want to talk about tailoring website search functionality for Chinese and Japanese languages. When it comes to entering “the East”, companies often face many challenges they could not have experienced before. Everything is different in China and Japan including the way how websites are built and how the users interact with them. In this article, I will cover one aspect of these challenges: how to adapt product/content search to work with Japanese and Chinese languages.

When we think about search engines or recommender systems, the default measure of quality is often relevance: does the system return what I asked for? Yet over time, it has become clear that accuracy alone does not create confidence. These systems don’t just retrieve information—they curate visibility, shape opportunity, and implicitly set the terms of what users come to rely on. That’s why the discussion has shifted toward a broader question: can we trust the ranking we see?