A paradigm shift in retail search
Over the last two decades, retail search has made quite a journey from its roots in full text and keyword matching search engines to the sophisticated semantic and visual search platforms of today.
To meet customer demands for relevant, timely and personalized search results, semantic search employs a wide array of techniques such as semantic query parsing, query expansion with domain-specific knowledge graphs, intent classification, deep learning-based natural language processing and “semantic” modeling of catalog data and customer’s shopping behavior. These techniques make the product discovery experience frictionless and delightful, with direct impact on conversion rates and revenue per session.
Semantic vector search is a powerful self-learning product discovery system that can be trained to achieve explicit business goals, such as click-through ratio or order conversion. By training a deep learning model based on all available catalog data and customer engagement history mined from the clickstream (attributes, images, descriptions and reviews, prices and promotions, etc.), semantic vector search captures the latent features (additional/hidden data) of queries and products and represents them as semantic vectors in a multidimensional vector space. Semantically similar products and queries are clustered together in the vector space, where nearest neighbors are ranked as the most relevant matches.
How semantic vector search improves the relevance of search results
By mapping concepts that are similar in meaning to be close in terms of vector distance, semantic vector search is primed for making sense of long-tail queries such as out-of-vocabulary natural language queries, thematic and subjective queries, and also directly addresses the central problems of polysemy and synonymy.
Thematic or subjective queries: Based on rich and detailed catalog data, semantic vector search can directly tackle thematic searches, when customers are vaguely or subjectively referring to a particular class of products, such as “chic dress” or “luxurious watch”. Symptom searches, when customers are referring to a problem rather than a product that solves it, can also be easily supported by semantic vector search based on analysis of customer shopping history.
Polysemy: Polysemy means that the same word may represent different concepts, like the word “table” in “table game” vs “accent table”. When it comes to polysemic queries, semantic vector search easily resolves ambiguity of alternative meanings based on customer behavior.
Synonymy: Synonymy means that the same concept can be represented in many ways by different words, like “Big Apple”, “NYC” and “New York” which all refer to the same city. Semantic vector models support basic synonymy with the help of NLP models which are pre-trained on a large volume of e-commerce data. On top of that, terms used by customers can be associated with relevant products when the semantic search model is trained on customer engagement data.
Key benefits of semantic vector search
Better query understanding
• Self-learning close-loop model
• Based on customer engagement data
• Meaning-based query interpretation
• Synonyms, spelling, slang and abbreviations
Better result ranking
•Self-learning close-loop ranker
•Supports business metrics (sales, margins…)
•Optimized for business goals
•Seamlessly supports personalization
Better understanding of the catalog
•Employs all available catalog data
•Titles, attributes, descriptions, reviews
•Price & promotion
•Image features, style extraction
Value-add use cases
•“Related queries” recommendations
•“Related products” recommendations
•Auto-complete
•Facet selection and ranking
Our blueprint for building a semantic vector search engine
Read more
Get in touch
Let's connect! How can we reach you?
Thank you!
It is very important to be in touch with you.
We will get back to you soon. Have a great day!
Something went wrong...
There are possible difficulties with connection or other issues.
Please try again after some time.