The detection of the bloggers with multiple weblogs
碩士 === 明新科技大學 === 資訊管理研究所 === 101 === Weblog receives increasing attention owing to the personal nature and features integration of Web 2.0 service. Bloggers construct their personal weblogs and cluster together because the similar interests, which are called online community. Conventional social ne...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Online Access: | http://ndltd.ncl.edu.tw/handle/41403879618752684067 |
| Summary: | 碩士 === 明新科技大學 === 資訊管理研究所 === 101 === Weblog receives increasing attention owing to the personal nature and features integration of Web 2.0 service. Bloggers construct their personal weblogs and cluster together because the similar interests, which are called online community. Conventional social network analysis of online community treats each weblog as an independent node on the internet. However, this fashion leads into error because a single blogger may contain multiple weblogs. Therefore, to construct precise online communities among heterogeneous blogospheres requires the identification of the bloggers who have multiple weblogs.
The profile in the weblog contains sufficient information which is suitable for blogger identification. This research surveyed the Wretch and Yam blogospheres to identify the bloggers with multiple weblogs. The identification was based on the accounts and profiles. This research adapted Edit Distance and Jaro Winkler Distance to compare the accounts. The profile was extracted into different attributes and computed the similarity of each attribute between the weblogs. The possibility of a blogger with multiple weblogs was estimated by combining the similarity of account and profile.
The experiment showed that the synthetic comparison of both the account and the profile outperforms than the only account identification. The experiment also revealed that the identification of the blogger with multiple weblogs reached 95% precision when the account similarity relaxes to 70%. This research can be applied in elevating the accuracy of the social network combination. This research also benefits in the name matching problem in heterogeneous social network services and crime prevention, like detect similar internet scam group or phishing websites.
|
|---|