improve efficiency of basic WordPress search for finding WooCommerce products by symbolic-numeric product attribute values

Hello there! We have a wp/woocommerce site with products having lots of custom (non-global) attributes, which are assigned via csv import. You can check the site here: [login to view URL] Attributes can be found under "Cross Reference" product tab, as well as some other tabs. As you can see, attribute values are primarily symbolic-numeric identifiers like "17220-PP4-E00". The problem: we want to have our search as smart and relevant as possible regardless of the value formatting (symbols, spaces). For example: 1. user searches for "17220-PP4-E00", he should be able to find products with attribute values of "[login to view URL]", "17220PP4E00" 2. user searches for "17220PP4E00", he should be able to find products with attribute values of "17220PP4 E00", "17220/PP4/E00" Results within a single attribute should be considered more relevant than ones across multiple attributes (eg attribute1: 17220, attribute2: PP, attribute3: 4, etc) We currently use Relevanssi plugin indexing attribute values, and a minor extension to replace nonalphanumebric symbols in the search query with spaces. That allows us to get some results, but relevance is still poor and example 2. is not possible. Keep in mind, that woocommerce keeps all nonglobal atributes and corresponding values serialized within a single product custom field. Requirements for bidders: you must have experience with search engines/systems you must specify in your proposal how exacly are you planning to solve the problem you are free to use any lib/plugin, not necessarily Relevanssi, but keep in mind that the search should still function for products as well, consider products more relevant than attributes, and keep other Relevanssi features.