We issued the same query (i.e, "ontology engineering")on the two datasets order to compare the results. Dirk and Claudia have the same interest for ontologies but the Dirk desktop contains less data than Claudias. Table 2 shows the top 5 results on the two datasets. We can see that the results are similar as Dirk and Claudia also share some publications on their desktops. While all the top 5 retrieved people have been working on the topic, the ranking might be improved. For example, the candidate Dirk Wenke"has less experience than “ Nicola guarino”or“ Rudi studer” on the topic. The explanation of this result is that only local evidence of expertise is used. The quality might be improved by looking at evidence on the web(e.g, DBLPS pages) Dirk Claudia Steffen Staab Steffen Staab 2 York Sure Riichiro Mizoguch Rudi Studer Dirk Wenke Dirk wenke York Sure 5 Nicola Guarino Rudi Studer Table 2. Top 5 results for the query "ontology engineering On Dirk's data we issues the query "pagerank"meaning the famous link ased algorithm proposed by Brin and Page in 7. The top 5 results are pre- sented in Table 3. We can see, again, that all the retrieved candidates have some experience on the topic, but the ordering is not good enough. The authors of the algorithm are placed fourth and fifth while they should be at the top of the list. The first three retrieved candidates have been working on the P2P version of the algorithm. Claudia Claudia ranking in information retrieval document search 1 Karthikeyan Sankaralingam 2 Simha Sethumadhavan Karl aberer Karl Aberer 3 James C.Browne Lawrence Pag Eli Upfal 4 Sergey brin Jon Kleinberg Sergey brin Lawrence Page Eli Upfal Monika Henzinger Table 3. Top 5 results for the query "pagerank"on Dirks desktop. Top 5 results for the queries"ranking in information retrieval""document search"on Claudias desktop. The query "religion"on Dirk's desktop, as expected, returned no results This can be explained because there is no evidence of expertise on such topic in this dataset http://www.informatik.uni-trier.de/-ley/db/We issued the same query (i.e., “ontology engineering”) on the two datasets in order to compare the results. Dirk and Claudia have the same interest for ontologies but the Dirk desktop contains less data than Claudia’s. Table 2 shows the top 5 results on the two datasets. We can see that the results are similar as Dirk and Claudia also share some publications on their desktops. While all the top 5 retrieved people have been working on the topic, the ranking might be improved. For example, the candidate “Dirk Wenke” has less experience than “Nicola Guarino” or “Rudi Studer” on the topic. The explanation of this result is that only local evidence of expertise is used. The quality might be improved by looking at evidence on the web (e.g., DBLP8 pages). Dirk Claudia 1 Steffen Staab Steffen Staab 2 York Sure Riichiro Mizoguchi 3 Rudi Studer Dirk Wenke 4 Dirk Wenke York Sure 5 Nicola Guarino Rudi Studer Table 2. Top 5 results for the query “ontology engineering”. On Dirk’s data we issues the query “pagerank” meaning the famous link based algorithm proposed by Brin and Page in [7]. The top 5 results are pre￾sented in Table 3. We can see, again, that all the retrieved candidates have some experience on the topic, but the ordering is not good enough. The authors of the algorithm are placed fourth and fifth while they should be at the top of the list. The first three retrieved candidates have been working on the P2P version of the algorithm. Dirk Claudia Claudia pagerank ranking in information retrieval document search 1 Karthikeyan Sankaralingam Sergey Brin Jon Kleinberg 2 Simha Sethumadhavan Karl Aberer Karl Aberer 3 James C. Browne Lawrence Page Eli Upfal 4 Sergey Brin Jon Kleinberg Sergey Brin 5 Lawrence Page Eli Upfal Monika Henzinger Table 3. Top 5 results for the query “pagerank” on Dirk’s desktop. Top 5 results for the queries “ranking in information retrieval” and “document search” on Claudia’s desktop. The query “religion” on Dirk’s desktop, as expected, returned no results. This can be explained because there is no evidence of expertise on such topic in this dataset. 8 http://www.informatik.uni-trier.de/~ley/db/