This paper presents the approaches proposed by the IRIS team of the IRITlaboratory for the TREC Incident Streams and Complex Answer Retrieval tracks. TheIncident Streams (IS) track aims to categorize and prioritize tweets in disaster situationto assist emergency service operators. In our participation, we applied supervised learningtechniques based on features extracted from tweets.Then, the 2019 edition of the Complex Answer Retrieval (CAR) track aims to answer com-plex questions expressed as outlines using paragraphs from Wikipedia. In our participation,we used the Terrier retrieval system to rank paragraphs for each section of the outline andkeep the most relevant according to three different strategies.