Drowning in opinions: Extracting the pearls

International Conference on Big Data Analysis and Data Mining

May 04-05, 2015 Kentucky, USA

Bradley Meyer and Justin Zhan

Scientific Tracks Abstracts: J Comput Sci Syst Biol

Abstract :

Consumers frequently solicit the opinions of others prior to making their purchasing decisions. Opinions in the form of online product reviews are commonly found on most major shopping web sites. The large number of reviews can however be over-whelming. As such recommender systems are often employed to sort through the data in search of the pearls of data useful to the consumer. Part of many recommender systems, especially when working with unstructured data, is the ability to extract product aspects from the streams of data. Aspect identification and extraction is critical to answer how and by what is the product judged? In this paper we present a novel aspect extraction technique which combines Latent Dirichlet Allocation (LDA) topic modeling with Part Of Speech (POS) language models to identify opinions with high levels of sentiment and are thus likely to be more informative relative to their LDA only counterparts.

Biography :

Bradley Meyer has completed his Masters in Statistics from Columbia University, and Under-graduation in Statistics from Baruch College. He is a professional programmer in multiple high-level languages and scripting languages. He is currently a Computational Science PhD student concentrating in distributed high performance computing and big data technologies. His main thrust is to perform data analysis via machine learning in a Hadoop centric environment over 10?s terabytes of data or greater.