User# E-mails Publications Addressbooks Calendars 0 234 12456 0 62013 218 28 10101001111 90123 116 1767 686 452 0007 01000001000002 Total 29409 2101 Table 1. Resource distribution over the users in the L3S Desktop Data Collection. A total number of 48,068 desktop items(some of the users provided a dump of their desktop data, including all kinds of documents, not just e-mails, publi cations, address books or calendars) has been collected, representing 8IGB of data. On average, each user provided 3, 433 items. order to emulate a standard test collection, all participants provided a set of queries that reflects typical activities they would perform on their desktop In addition, each user was asked to contribute their activity logs, related to the period until the point at which the data were provided. All participants defined their own queries, related to their activities, and performed search over the reduced images of their desktops, as mentioned above 4 Preliminary Experiments We used the Dirk and Claudia datasets (see Section 3. 1)in order to perform some initial evaluation of our system for finding experts. We created some queries that match the personas interests imagining which kind of experts they would need to find The expert search queries on the Dirk's desktop are pagerank The expert search queries on the Claudias desktop are: ontology engineering ranking in information retrievalUser# E-mails Publications Addressbooks Calendars 1 109 0 1 0 2 12456 0 0 0 3 4532 1054 1 1 4 834 237 0 0 5 3890 261 1 0 6 2013 112 0 0 7 218 28 0 0 8 222 95 1 0 9 0 274 1 1 10 1035 31 1 0 11 1116 157 1 0 12 1767 2799 0 0 13 1168 686 0 0 14 49 452 0 0 Total 29409 6186 7 2 Avg 2101 442 0.5 0.1 Table 1. Resource distribution over the users in the L3S Desktop Data Collection. A total number of 48,068 desktop items (some of the users provided a dump of their desktop data, including all kinds of documents, not just e-mails, publi￾cations, address books or calendars) has been collected, representing 8.1GB of data. On average, each user provided 3,433 items. In order to emulate a standard test collection, all participants provided a set of queries that reflects typical activities they would perform on their desktop. In addition, each user was asked to contribute their activity logs, related to the period until the point at which the data were provided. All participants defined their own queries, related to their activities, and performed search over the reduced images of their desktops, as mentioned above. 4 Preliminary Experiments We used the Dirk and Claudia datasets (see Section 3.1) in order to perform some initial evaluation of our system for finding experts. We created some queries that match the personas interests imagining which kind of experts they would need to find. The expert search queries on the Dirk’s desktop are: – ontology engineering – pagerank – religion The expert search queries on the Claudia’s desktop are: – ontology engineering – ranking in information retrieval – document search