Inbenta can dissect a query into Base entities (or product type) and Attribute entities (or product properties). It leverages the Inbenta Lexicon, Entity Detection, and Classification capabilities to fully understand natural user queries.
|User Search:||“blue iPad cover under $100”|
|Inbenta Lexicon:||Blue (Adjective) + iPad (Noun) + Cover (Noun) + Under (Preposition) + Dollar (Noun) + 100 (Number)|
|Entity Detection / Classification:||Base Entities: iPad (Noun) + Cover (Noun)|
|Attribute Entities:||Color: Blue (Adjective)|
|Price Range:||Max = 100 (Number),
Min = 0,
Currency = Dollar (Noun)
|Base Entity Search:||returns all “iPad Cover”|
|Attribute Entity Search:||apply automatic filtering to narrow down the results for “iPad Cover”, and return all “blue” and “below $100” items.|
Every e-commerce business is different, and no standard solution can adapt to your particular needs. Inbenta is keen to work with you in order to find the best solution on how to adapt our AI and NLP to your particular needs. Smart search, based on artificial intelligence, natural language processing and a custom lexicon, really understands what to look for. It is enriched with the following features:
- Implicit Entity Recognition allows for context analysis to identify implicit elements not clearly defined by the user. For example, a user who searches for “1 1/2 nut” on a home improvement site will see Wing Nuts filtered by a 1.5-inch span.
- Negation Recognition detects when a user is negating or excluding product properties from their search. For example, searching for “wool sweaters that are not green” will return all wool sweaters filtered by every color except green.
- Abstract Qualifiers allows for advanced treatment of qualifiers that may change in meaning depending on context. For example, when searching for “small adapters,” the engine recognizes “small” as an abstract qualifier. It maps “small” to the lower portion of the “cord length” range which was implicitly inferred from the context. The definition of “small” can be defined differently depending on the product type or even the product category.
- Range Recognition enables the smart array recognition of product properties. For example, searching for “jeans $80 to $200” will return jeans filtered by price between $80-$200 inclusive, and searching for “1/2 to 2.5 film tapes” will return film tapes filtered by width between 1.5-2.5 inches inclusive.
- Unit of Measure (UOM) Conversion improves search relevance by automatically performing unit of measure conversions on the fly. For example, a search for “20-quart paint pail” will return 5-gallon paint buckets as well, converting quarts to gallons behind the scenes and identifying the proper match.
- Long tail SEO uses customer input to generate optimized landing pages that are easily crawled. When a different customer makes a similar search, they will be taken to the optimized landing page where they can buy what they are looking for. If people change the way they search for a product, the page will update accordingly so that customers can always find what they want.
Customers on an e-commerce site with efficient search will be up to 50% more likely to buy because they found what they want. When a visitor to your website always finds what they want right away, it builds customer trust and loyalty.