Actor Reputation
From SONIVIS:Wiki
Actor Reputation
|
Contents |
Objective
actor reputation
Explanation (optional)
Assuming an actors within a virtual information space can make changes to its content, actors gain reputation when the edit they preformed is not changed by others and lose reputation when their edits are removed[1]. This reputation is content-driven, since it is calculated without any user input.
Adler, T. B. and de Alfaro, L. 2007[1]: "Suppose that an author A contributes to a Wikipedia article by editing it. When another author B subsequently revises the same article, she may choose to preserve some of the edits performed by A. By preserving them, B provides her vote of confidence in these edits, and in author A. Our reputation system will increase the reputation of A in a manner that depends on the amount of preserved edits, as well as on the reputation of B. Specifically, our system measures the following quantities text life (How much of the text inserted by A is still present after B’s edit) and edit life (How much of the reorganization (text reordering and deletions) performed by A is preserved after B’s edit). Author A receives a reputation increment proportional to the size of his contribution, to the text and edit life of the contribution, and to the amount of reputation of B. Author A receives the largest reputation increment when B preserves his contribution in its entirety; conversely, A’s reputation is most negatively affected when B rolls back A’s contribution. More subtly, the way we compute text and edit life ensures that if B modifies the contribution of A, building upon it, then A still receives a reputation increment."
Adler and de Alfaro[1] say the reputation of the involved actors of a text can be used as rough guide to the veracity or trustworthiness of actor or content. A precise argumentation that revision information can be used to compute a measure of trustworthiness can be found in [1]. It can be used for author management, can be a incentive for authors to provide high-quality contributions, alert editors when a low-reputation actor makes changes, or can be used to predict the quality of author's contributions.
Still the content-reputation can be wrong. The content can be erased for many reasons (article rewrite, article movement...). To cope it one has to distinguish between reverted and refined contribution. "Authors of reverted contributions lose reputation, while authors of sebsequently refined contributions do not [...]. Second, contributions that are appropriate, but that are subsequently rewitten, tend to last longer than in appropriate contributions"[1].
Calculation
see [1]
Reference
@inproceedings{citeulike:1291537,
address = {New York, NY, USA},
author = {Adler, Thomas B. and de Alfaro, Luca },
booktitle = {WWW '07: Proceedings of the 16th international conference on World Wide Web},
doi = {http://dx.doi.org/10.1145/1242572.1242608},
keywords = {content-driven, reputation, wikipedia},
pages = {261--270},
publisher = {ACM Press},
title = {A content-driven reputation system for the wikipedia},
year = {2007}
}
@inproceedings{citeulike:2359429,
author = {Zeng, Honglei and Alhossaini, Maher and Ding, Li and Fikes, Richard and Mcguinness, Deborah L. },
keywords = {quality, wikipedia},
organization = {Proceedings of the 2006 International Conference on Privacy, Security and Trust},
title = {Computing Trust from Revision History},
url = {http://ebiquity.umbc.edu/paper/html/id/320/Computing-Trust-from-Revision-History},
year = {2006}
}
