Abstract | ||
---|---|---|
Finding potential customers in social networks is a hard challenge for today's businesses. But by listening to the noise of social network posts, we identify users, who express a demand for a certain product. We achieve this identification with a two-stage text categorization classifier: First, we detect whether the post expresses a demand for some product in general. Second, we detect, which product the post is about. By using the company's brochures, we minimize the integration effort for our system. However, this approach is difficult, because brochures differ from social network posts in style and length and only few brochures exist for each product. By employing feature selection and document sampling we are able to cope with these issues. Our evaluation has shown the practicability of this approach and supports our decisions for a two-stage classifier, document sampling and strict feature selection. |
Year | DOI | Venue |
---|---|---|
2015 | 10.1109/SmartCity.2015.109 | 2015 IEEE INTERNATIONAL CONFERENCE ON SMART CITY/SOCIALCOM/SUSTAINCOM (SMARTCITY) |
Field | DocType | Citations |
Customer relationship management,Data science,Social network,Social web,Feature selection,Computer science,Active listening,Software,Artificial intelligence,Sampling (statistics),Classifier (linguistics),Machine learning | Conference | 0 |
PageRank | References | Authors |
0.34 | 9 | 6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Philipp Berger | 1 | 17 | 8.14 |
Patrick Hennig | 2 | 14 | 7.38 |
Stefan Bunk | 3 | 0 | 0.34 |
Dimitri Korsch | 4 | 0 | 0.34 |
Daniel Kurzynski | 5 | 0 | 0.34 |
Christoph Meinel | 6 | 2341 | 319.90 |