Intelligent Product Lifecycle Management

Nowadays, consumers have the luxury of selecting many different types of products that are available on the market. As a matter of fact, the challenge for the consumers is not to find the product they need but to find the “right” one from the excessive, and sometimes overwhelming, choices and options available to them. Some believe that this abundance is a powerful indication of consumerism, an economic policy that emphasizes, and encourages, consumption. For example, the average household spends close to $1,500 on just consumer electronics every year with an annual approximate increase of 12%. With the economic challenges that many consumers (being individuals or larger corporations) face, especially in North America, the consumers are becoming more conscious of efficient expenditure and asset reuse. In order to be spending more prudently, consumers need to be equipped with information such as what products they own; the products’ features; warranty information; and resale value. As products are purchased over the years (from thousands of items in a household to millions in a corporation), such information may not be easily accessible when it comes time to make another purchase.

Warranty Life is a Web 2.0 Technology Company that builds the toolset for managing product purchases of consumers for the entire lifecycle of the product. The tools offered by Warranty Life manage consumers’ purchases, protect them with extended warranties, identify manufacturer supported warranties, repair them if required and offer an opportunity to resell, recycle, or replace the items. The entire framework of tools supports consumers in efficiently using and reusing the items they own and balancing their expenditure. In its current form, Warranty Life allows its users to enter information about the products that they own or recently purchased into the system using structured input forms or by submitting their electronic receipts to the Web site after which a human operator will input the information on behalf of the user; this information will become later accessible and searchable to/by the user. However, there are challenges that impede this process, including: 1) the numerosity of the items that each user (e.g., a large corporation) owns makes the initial information input process slow and tedious; and 2) the use of different terminologies to describe similar items and products (e.g., laptop and notebook) causes ambiguity in product description and inhibits efficient product search and retrieval.

To address these challenges, we undertake the following two major research activities:

  1. Information extraction from record documents – the first major step in our research plan is to investigate various methods and techniques for the extraction of useful pieces of information about the products that a user owns from proprietary record documents, e.g., sale receipts. This process will include the use of natural language processing techniques such as named entity recognition to find valuable data about the product; this way, the user can scan any documents that he/she has about a product and that will be automatically inserted into the private product knowledge base of the user – reducing the information input problem mentioned earlier. The long-term vision is to allow a user to take a photo of any product record using his/her smart phone and it will be automatically processed for product information and stored in the product knowledge base. Given the scope of this project, the assumption is that all product record documents are represented in machine processable text documents, i.e., the step for converting image documents into text documents that constitutes image processing is outside the scope of this project;
  2. Semantic product representation– the second focus of this project is on the exploration of foundations for representing product information in such a way that semantics and meaning are preserved when information are stored. As mentioned earlier, product information may be stored using different syntactical structures (e.g. synonyms). In such cases the use of a shared semantic representation for products would allow for a more accurate and precise search and retrieval process. In addition, there are cases when a search for a product does not necessarily need an exact term match but rather the user is searching for products that have similar functionality. In such cases, a semantic representation of the products along with their features, attributes, and functionalities would enable the system to perform semantic reasoning on the information in the knowledge base and retrieve otherwise inaccessible information. An example would be when a company is deciding whether to buy several iPad devices to serve as ebook readers. A product search for iPad will not return any results and hence encourage the company to buy these devices, while the company already owns Kindle devices that would have been returned as a result of search if semantic product representation and inferencing were employed. We are interested in ontological product representations in OWL (e.g., the LOD cloud) for knowledge representation and Description Logic reasoners such as Pellet and FaCT++ as reasoning engines.


Natural Sciences and Engineering
Research Council of Canada