Name
Abstract | ||
|---|---|---|
In our former research, customer's preference has been estimated by passive observation of shopping behavior, e.g. customer's "look" and "touch". It takes much time to understand their preferences form the log. We need quickly to build up the preference model to perform suitable recommendation for a new customer. For this reason, we will propose an active observation mechanism that detects customer's unforced natural behavior to information through ambient devices such as speakers and electric displays. This mechanism also analyzes customer's preference on features and their values of commodities, which enables the system to estimate the rate of preference to an unknown product. We have experimented on ten university students. We had them evaluate the thirty-six Shirts. We used these evaluations for precision evaluations in naive Bayes classifier. We used the leave-one-out cross-validation. As the result, we have achieved the average precision in the estimating preferences by naive Bayes classifier is 71%. |
Year | DOI | Venue |
|---|---|---|
2013 | 10.1007/978-3-642-39265-8_52 | HCI (3) |
Keywords | Field | DocType |
new customer,detects customer,unforced natural passive observation,precision evaluation,naive bayes classifier,active observation mechanism,passive observation,analyzes customer,shopping behavior,preference model,average precision,customer preference,recommendation system | Recommender system,Customer preference,Naive Bayes classifier,Computer science,Digital signage,Human–computer interaction | Conference |
Citations | PageRank | References |
1 | 0.35 | 1 |
Authors | ||
3 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Terumasa Tajima | 1 | 1 | 0.35 |
| Yusuke Iida | 2 | 4 | 1.01 |
| Toshikazu Kato | 3 | 1 | 0.35 |