@jaymyers’ talk from Best Buy
Product attributes are valuable data, and it’s the secret behind the passion of Linked Data that Jay is presenting on.
The goal (personal) is to “provide more visibility to products, services, and locations to humans and to machines.”
The good news, for retailors, is that the tools are already available!
RDF(a), Microformats, GoodRelations, and HTML
Brick and mortar stores (Best Buy)
Location info (i.e. where the store is) is locked away, it’s siloed. So, the first step was to publish a page for every store. You can think of each store (or page about the store) as a bundle of interesting data (where the store is, what’s in stock, when’s it open?)
The next step was to change the pages into individual blogs. Each store has two people who can update the blog with very simple data, like changes to opening times. Behind the scenes, they’re publishing metadata about the store, as RDFa in GoodRelations.
Big, unintentional win, being a huge increase in search traffic without any intentional keyword or other traditional SEO investment. Sounds, to me, a lot like the results BBC’s wildlife finder found.
Open Box Products (i.e. returned, fully-functioning goods)
The stores can be seen as a little data silo, holding information on how many, of which products are being returned.
This costs the industry billions.
So, the store employees were given a form with the SKU number of the returned product, and a few fields for what the discount was (i.e. how much it should have been, and by how much it was discounted), as well as the reason for its return.
This is published as RDFa, and they’re seeing the beginning of a relationship between products, returns, locations and reasons emerging.
- SEO and product visibility (semantics are hugely useful for these)
- Reduction of proprietary data feeds.
One of the good practices of the Semantic Web is to reuse ontologies to name common concepts and move towards a state where different domains are interlinked both through common namespaces and references between the data.
In case of Idea Management this could lead to easier integration with other enterprise management systems such as PLM, PLCM, ERP etc.; development and engineering tools (e.g. bug tracking systems, IDEs); social spaces (popular portals such as Twitter or Facebook). GI2MO aims to follow this trend, however within reason. To create and apply the ontology we constructed a generic data model for the Idea Management system based on doamin study and selected a number of concepts to model with Idea Management domain ontology and a limited number with other.
Idea Life Cycle and Actors
The modern Idea Management Systems attempt to build their workflow around the concept of Idea Management as a repeatable process. To achieve that Idea Life Cycle has been defined and its phases have been specified to formalize the domain. However, depending on the product or research project those cycles sometimes have slight differences, both in naming and cycles amount or definition. In our research we have assumed the following organization and notation of the Idea Life Cycle:
- Idea Generation
- Idea Improvement
- Idea Selection
- Idea Implementation
- Idea Deployment
Each of the phases can defines different actors and operations (see Fig. 1) that in turn produce different data that enriches idea and connected assets in different ways.
Fig.1 Idea Life Cycle and actors involved in the process.
3. Reference Idea Management System data model
Fig.2 GI2MO Data Model Overview
- Why don’t you follow the Semantic Web practice? Well, it depends on how you define a good practice @ Been Accountable (foolme.csail.mit.edu)
- An Executive Intro to Ontologies (mkbergman.com)
- Tutorial 3: Semantic Modeling (linkeddatatools.com)
- Case Study: Enhancement and Integration of Corporate Social Software Using the Semantic Web (w3.org)
- Interlinking the social web with semantics (laurensgoessemantic.wordpress.com)