The Product Graph Company
Becoming the “The Product Graph Company” requires building (reverse engineering) a database of the products at each product site on the Internet. Matched product records which are the same product found at different sites, including variants. Detecting and eliminating bad data is part of the matched record creation process. Facebook is the social graph company. Google is the knowledge graph company.
Which company is the product graph company? Some would argue that Amazon is the product graph company. But are they really? After analyzing the Amazon product data it becomes very clear that Amazon does not do a good job de-duplicating product nor do they do a good job eliminating variants nor do they do a good job of idenifying and fixing bad and missing data. Moreover, their product matching leaves a lot to be desired. In addition Amazon does not have a complete database of all product data on the web.
Some would say that Google has done a good job at extracting structured product data and converting product records in data feeds uploaded by merchants. Google product data search results are incomplete. Search by variants such as size, color, and material is not possible. Search results start with paid placement ads. And finally Google does not have complete information for products nor does Google have a complete matched record for all products.
Pinterest is a social graph. And Pinterest contains products. Surely Pinterest must be the product graph everyone is looking for. Alas Pinterest does not have structured information for non-paid pins. And paid pins only have a limited set of structured data sent by the store which has the data. Pinterest is developing a “you might also like” image similarity feature and a feature to show ads with images which are similar to the images a user is looking at. At the end of the day Pinterest is not the product graph company.
Facebook is a huge company and generates a lot of its revenues from brand advertising. Facebook bought Thefind which was a shopping engine. Facebook must have deep insights into product information. After all Facebook is heavily into machine learning and AI. Facebook offers ads in your newsfeed. That is it. There is no shopping platform on Facebook. Product information does nto seem to be used to any large degree on Facebook.
Twitter should be a product graph company. We had a patent granted for analyzing sentiments about products, brands, and stores. Twitter is sitting on a goldmine of user data that can be used for customer service, martketing, predictive analytics, and other applications. I have heard rumors that they are building something in this area but I have seen nothing concreete to date.
“13. The method of claim 1 wherein the semantic analysis detects user sentiment on one or more token groups selected from the hierarchical structure of token groups in the social information records, consisting of brand attributes, product features, store and brand policies, service, durability of the product, suitability of the product, longevity of the product, design of the product, comparison of the product or brand to other brands, performance, problem, deal, purchase, question, recommendation, satisfaction, value, wish, design, specification, construction, customer service.”
Several companies have attempted to create matched product record databases. However, these companies publicly acknowledge that their DB’s have limitations.
Who is going to be the product graph company?