Abstract
Fast and accurate search results are a crucial components of any e-shop and thus can make the difference between high user satisfaction and user frustration. With recent advancements in vector search technologies, enhanced search systems have become more efficient, leading to better user experiences and improved conversion rates. In this talk, we’ll explore how to implement a hybrid search system for a non-english e-commerce site using Pinecone, a high-performance vector search engine.
Prerequisites
Knowledge of natural language processing basics (tokenization, embeddings) are preferred, but not required.
Description
Vector search has unlocked the door to another level of relevance and efficiency in information retrieval, but it isn’t a perfect technology. In fact, without big domain-specific datasets to fine-tune models on, a traditional keyword search still has some advantages. We repeatedly see that vector search unlocks incredible and intelligent retrieval but struggles to adapt to new domains or deal with industry-specific jargon as well as rare words. On the other hand, traditional search can cope with new domains but is fundamentally limited to a set performance level and is unable to capture semantic relationships within the query. As we can see, both approaches have pros and cons, but what if we merge them somehow to eliminate a few of those cons? Could we create a hybrid search with the heightened performance potential of vector search and the zero-shot adaptability of traditional search? In this talk, we will take both vector and traditional search and merge them via Pinecone’s new hybrid search to create a flexible and friendly search engine, capable of understanding queries of diverse nature. By the end of this talk, you’ll not only have a strong understanding of hybrid search but also a solid intuition on how to implement it, improving the search functionality of any e-commerce site.
