Abstract. This paper tackles the problem of pinpointing relevant information in a social network for Personalized Information Retrieval (PIR). We start from the premise that user profiles must be filtered so that they outperform non profile based queries. The formal Profile Query Expansion Constraint is then defined. We fix a specific integration of profile and a probabilistic matching framework that fits into the constraint defined. Experiments are conducted on the Bibsonomy corpus. Our findings show that even simple profile adaptation using query is effective for Personalized Information Retrieval.