The XSearch Engine composer helps you configure search engines that leverage the advanced AI-driven models in XSearch Models for highly customized and relevant search experiences. With a focus on semantic behavior, users can choose from various model sets like DeepSearch, GenSearch, and Hybrid Search to optimize how queries are handled. The composer offers extensive options for query settings, including spell correction, negative tone handling, and explicit search. Users can fine-tune search results with thresholds, weighting, and translation settings, making it a versatile solution for various search use cases.
Model Selection
The first key setting in configuring an Engine is determining which model to use. The XGen Team will advise you on your selection, but we have provided a brief description of each below.
Search Model Options
There are four key models available:
DeepSearch: A highly sophisticated language-based search methodology that leverages a series of text-based models to find matches based on product data. DeepSearch is the best application for single-word and highly literal searches.
GenSearch: A generative search methodology that incorporates custom LLM-training and image-based search to infer meaning and context to natural-language queries and provide hyper-relevant results. GenSearch is the best application for complex, multi-word queries.
DeepSearch + GenSearch: A combine methodology that leverages the strength of DeepSearch and GenSearch in sequence, based on the word or character count input by the user.
Hybrid Search: Hybrid Search is an extension of GenSearch that incorporates additional product metadata into the training dataset, allowing for a blend of the two search methodologies.
