Abstract | ||
---|---|---|
In this work, we are proposing several approaches to enhance lost/won classification of complex deals using sentiment analysis. The analysis of sentiments is done by text mining the activity notes recorded in CRM Systems used to manage complex sales. Using a baseline SVM model, we extended the baseline features with opinion predictors gathered using various techniques that included different preprocessing approaches of the CRM notes, scoring and counting of opinion sentences and inference of sentiment level features. We analyzed and compared the accuracy and f1-measure gained in comparison to the baseline and we discovered that, among the approaches analyzed, counting the polarity sentences gives the highest gain. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1109/SYNASC.2017.00038 | 2017 19th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC) |
Keywords | Field | DocType |
Customer Relationship Management,Classification,Opinion Mining,Support Vector Machines,Sentiment Analysis | Customer relationship management,Text mining,Inference,Sentiment analysis,Computer science,Support vector machine,Feature extraction,Theoretical computer science,Preprocessor,Natural language processing,Artificial intelligence | Conference |
ISSN | ISBN | Citations |
2470-8801 | 978-1-5386-2627-6 | 0 |
PageRank | References | Authors |
0.34 | 9 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Doru Rotovei | 1 | 0 | 1.69 |
Viorel Negru | 2 | 311 | 47.71 |