Zu Hauptinhalt springen

Lehrstuhl für Wirtschaftsinformatik III

Projekte2


UR:SMART

How can customer opinions get extracted from social media channels automatically and analyzed to get deep insights into current needs?

UR SMART (University of Regensburg Social Media Analysis Research Toolkit) is a tool to extract, analyze and interpret social media posts, which is adapted to the particular needs of small and medium-sized enterprises (SMEs) in Bavaria. Via a graphical user interface (GUI) employees may easily analyze customer posts on a company’s Facebook page or the Twitter channel(s) to gain deep insights into consumers’ current attitudes and needs. For that purpose, UR SMART offers an automatized sentiment analysis and a topic-related classification of posts. UR SMART was developed in close cooperation with three partners from industry.

Smart1


Smart2

Sentiment-Analysis

The sentiment analysis is performed. Depending on the sentiment of each word, the annotated sentiment score of a post is either positive, neutral or negative. It is expressed by a number within a predefined range [-2,+2].

-->


Smart3

Classification of Posts

Social media content is assigned to predefined classes to identify those topics, the customers vividly discuss in a positive, neutral of negative way. Based on that, management decisions can be derived straightforward.

-->


Smart4

Analysis of time series

The impact of certain events (e.g. presentation of new products) can be observed by a graphical illustration of the sentiment along the observed time series.

-->


Smart5

Combination of Sentiment and Classification

Every post is analyzed in parallel for the sentiment as well as for the most appropriate class. By combining both findings, the need for action can be derived. E.g. extraordinary negative posts for a certain product can be an advice to redesign that product.

-->


Score

Quantitative Measures

Besides the qualitative analysis regarding the sentiment or the posts` category, quantitative measures are also provided.

-->


Highlighting of notably posts

Notably positive or negative posts are presented separately to signal the need for action.

?

?

For Demo-Access please contact us via email:

florian.johannsen@wiwi.uni-regensburg.de

-->


  1. STARTSEITE UR
  2. Fakult?t für Informatik und Data Science

Lehrstuhl für Wirtschaftsinformatik III

Prof. Dr. Susanne Leist 

logo

Sekretariat:

Fax  0941 943-81-3201

E-Mail sekretariat.leist@ur.de