Title: AN IN-MEMORY APPROACH TO SENTIMENT ANALYSIS BASED ON SAP HANA

Issue Number: Vol. 5, No. 1
Year of Publication: 2015
Page Numbers: 1-13
Authors: Karl Kurbel, Dawid Nowak, Florian Jatzold, Pavlo Glushko
Journal Name: International Journal of Digital Information and Wireless Communications (IJDIWC)
- Hong Kong
DOI:  http://dx.doi.org/10.17781/P001505

Abstract:


Sentiment analysis (SA) is one of a number of fields in which "big data" requires processing. The term "sentiment" refers to the feelings, beliefs, emotions, and opinions that people express in relation to certain matters in written form, more often than not in online media such as social networks. Sentiment analysis can provide enterprises with valuable insights into, for example, how customers regard their products. When sentiments are collected from social networks over a period of time, the amount of data can indeed be "big". This paper describes and evaluates a lexicon-based SA solution, implemented as a prototype using SAP HANA. Sentiments are collected from Facebook and Google+, using a combination of tools including Microsoft Excel and the Power Query add-on to Excel. As HANA employs in-memory database technology, the calculation of sentiment values will be very fast. As a result of the sentiment analysis, five categories (from strong negative to strong positive), plus a category "sentiment not found", are filled with individual sentiments and, with the help of business intelligence tools, evaluated as well. SA results are made available on a user interface created using SAP's current interface technology, SAPUI5.