For my Senior Design project in Penn Engineering, a couple M&Ts and I (Markus and Krishna) are attempting to create a better way to search for products. Currently, when searching for a product, a rational customer must look at product specifications and match them to their end uses. This could involve considerable research by consumers on how product specifications relate to use. For example, when searching for a laptop, a user would have to match specifications such as graphics card and processor against their projected needs for the computer, such as gaming or video editing. Our search engine will build a layer of abstraction over product specifications so users can search directly by their end use.
Our project is attempting to achieve this goal by using publicly available review databases, for example, Amazon, Cnet, and Newegg. The abstraction layer is being built by applying statistical methods to the review dataset. We are exploring a variety of methods ranging from machine learning algorithms to a modification of the PageRank algorithm. Natural Language Processing has also proved to be a very challenging aspect of our project; one where we have had to devise some original solutions. I look forward to continuing our work and research into this project!