cuss the last two queries on Claudia's dataset. We created queries on very similar topics (i.e, "ranking in information retrieval"and"doc ument search") in order to compare the results. The results are shown in Table 3. We can see that the top 5 results are similar but the ranking. In this case it is hard to say which the best ranking should be as all the retrieved candidates have strong experience on the topic and deciding who is the most expert is highly subjective In conclusion, we have seen that the effectiveness of finding experts using the desktop content highly depends on the available resources. If the user queries for experts on topics well represented on her desktop, then the results can be satisfactory. If the query is off-topic then the results can be poor or even be missing. Moreover, further improvements are needed on the ranking function used. A novel measure replacing the cosine similarity used in this experiment might be used 5 Discussion of related work In this section we describe and discuss the previous work in the field of Expert Search and PIM. We show how existing systems have been designed, which formal models have been proposed, which PIM systems can be extended with expert search functionalitie 5.1 Expert Search Systems Several expert search systems have been proposed in the last years. These sys- tems use different information sources and features like social network informa- tion 9, co-occurrences of terms and changes in the competencies of people over time 30, rule-based models and FOAF data 21]. For the web, a different con- text from the enterprise search one, one of the approaches proposed 29 focuses on scenarios like Java Online Communities where experts help newcomers or collaborate with each other. and investigated several algorithms that build on answer-reply interaction patterns, using PageRank and HITS authority models as well as additional algorithms exploiting link information in this context. We are not aware of any system for finding experts on the desktop The Enterprise People Finder [ 25, 26 also known as P@noptic Expert [1 first builds a candidate profile attaching all documents related to that candidate in one big document giving different weights to the documents based on their An interesting distinction has been made between expert finding and expert profiling in 4. The former approach aims at first retrieving the documents relevant to the query and then extract the experts from them. The latter first builds a profile for each candidate and then matches the query with the profiles without considering the documents anymore 5 http://www.foaf-project.orgFinally, we discuss the last two queries on Claudia’s dataset. We created queries on very similar topics (i.e., “ranking in information retrieval” and “doc￾ument search”) in order to compare the results. The results are shown in Table 3. We can see that the top 5 results are similar but the ranking. In this case it is hard to say which the best ranking should be as all the retrieved candidates have strong experience on the topic and deciding who is the most expert is highly subjective. In conclusion, we have seen that the effectiveness of finding experts using the desktop content highly depends on the available resources. If the user queries for experts on topics well represented on her desktop, then the results can be satisfactory. If the query is off-topic then the results can be poor or even be missing. Moreover, further improvements are needed on the ranking function used. A novel measure replacing the cosine similarity used in this experiments might be used. 5 Discussion of Related Work In this section we describe and discuss the previous work in the field of Expert Search and PIM. We show how existing systems have been designed, which formal models have been proposed, which PIM systems can be extended with expert search functionalities. 5.1 Expert Search Systems Several expert search systems have been proposed in the last years. These sys￾tems use different information sources and features like social network informa￾tion [9], co-occurrences of terms and changes in the competencies of people over time [30], rule-based models and FOAF9 data [21]. For the web, a different con￾text from the enterprise search one, one of the approaches proposed [29] focuses on scenarios like Java Online Communities where experts help newcomers or collaborate with each other, and investigated several algorithms that build on answer-reply interaction patterns, using PageRank and HITS authority models as well as additional algorithms exploiting link information in this context. We are not aware of any system for finding experts on the desktop. The Enterprise PeopleFinder [25, 26] also known as P@noptic Expert [17] first builds a candidate profile attaching all documents related to that candidate in one big document giving different weights to the documents based on their type. An interesting distinction has been made between expert finding and expert profiling in [4]. The former approach aims at first retrieving the documents relevant to the query and then extract the experts from them. The latter first builds a profile for each candidate and then matches the query with the profiles without considering the documents anymore [5]. 9 http://www.foaf-project.org