Elicitability Export

@inproceedings{citeulike:3026076, abstract = {We investigate the problem of truthfully eliciting an expert's assessment of a property of a probability distribution, where a property is any real-valued function of the distribution like mean or variance. We show that not all properties are elicitable; for example, the mean is elicitable and the variance is not. For those that are elicitable, we provide a representation theorem characterizing all payment (or ” score”) functions that induce truthful revelation. We also consider the elicitation of sets of properties. We then observe that properties can always be inferred from sets of elicitable properties. This naturally suggests the concept of elicitation complexity ; the elicitation complexity of property is the minimal size of such a set implying the property. Finally we discuss applications to prediction markets.}, author = {Lambert, Nicolas and Pennock, David and Shoham, Yoav}, booktitle = {Proceedings of the ACM Conference on Electronic Commerce}, citeulike-article-id = {3026076}, citeulike-linkout-0 = {http://robotics.stanford.edu/~shoham/www%20papers/LambertEtAlEC08elicitability.pdf}, comment = {This paper generalises the scoring rule weighted integral representation results of Schervish (as discussed in Buja et al. 2005) to the elicitation of properties of distributions. They characterise which properties are "elicitable" - i.e., can be extracted from someone via a scoring rule - and show that these are precisely the properties that, for all values of the property, have convex level sets in the space of distributions. Very nice paper! }, keywords = {integral\_representation, prediction\_markets, probability, scoring\_rules, weighting}, month = {July}, posted-at = {2008-07-22 06:13:12}, priority = {0}, title = {Elicitability}, url = {http://robotics.stanford.edu/~shoham/www%20papers/LambertEtAlEC08elicitability.pdf}, year = {2008} }

TY - CONF ID - citeulike:3026076 L3 - citeulike-article-id:3026076 N2 - We investigate the problem of truthfully eliciting an expert’s assessment of a property of a probability distribution, where a property is any real-valued function of the distribution like mean or variance. We show that not all properties are elicitable; for example, the mean is elicitable and the variance is not. For those that are elicitable, we provide a representation theorem characterizing all payment (or “score”) functions that induce truthful revelation. We also consider the elicitation of sets of properties. We then observe that properties can always be inferred from sets of elicitable properties. This naturally suggests the concept of elicitation complexity ; the elicitation complexity of property is the minimal size of such a set implying the property. Finally we discuss applications to prediction markets. TI - Elicitability T2 - Proceedings of the ACM Conference on Electronic Commerce KW - integral_representation KW - prediction_markets KW - probability KW - scoring_rules KW - weighting N1 - This paper generalises the scoring rule weighted integral representation results of Schervish (as discussed in Buja et al. 2005) to the elicitation of properties of distributions. They characterise which properties are "elicitable" - i.e., can be extracted from someone via a scoring rule - and show that these are precisely the properties that, for all values of the property, have convex level sets in the space of distributions. Very nice paper! AU - Lambert, Nicolas AU - Pennock, David AU - Shoham, Yoav PY - 2008/07// UR - http://robotics.stanford.edu/~shoham/www%20papers/LambertEtAlEC08elicitability.pdf ER -