Reviewer A: Originality: How would you rate the originality of the paper?: Accept Relevance: How would you rate the relevance of this paper to the conference topics?: Accept Technical Quality: How would you rate the technical quality of this paper?: Accept Presentation: How would you rate the presentation of this paper?: Accept Guidance for Authors: Please describe in detail main paper contributions, positive aspects, observed deficiencies, and suggestions on how to improve them:: This paper proposes a methodology to infer land use patterns by fusing GPS traces from floating cars with mobile phone data. The main contributions of the paper are twofold: 1) Development of multi-domain signature based clustering for land use detection that utilizes multiple data sources. This work generalizes a previous approach of Furno et al, 2016 by refining the more coarse grained CDR-based land uses using the observed dynamics of taxis and floating vehicles. 2) The proposed methodology is applied and evaluated using large data sets from the cities of Lyon, Milan, and Naples, and the results suggest that the taxi/FCD data can be used to improve the land-use characterization obtained from solely CDR data, particularly in areas where CDR data is coarse. The paper is very well written, particularly the description of the methodology to construct the representative set of multi-metric signatures. The data sets and processing is adequately described and the conclusions are properly supported. Comments • The contribution in terms of the extension/generalization of previous work should be more clearly and explicitly described. From my understanding, the key contribution lies in the use of the additional taxi/FCD data, this needs to be clearly stated at the outset. Questions • Was any sensitivity analysis done on the 4 hour aggregation interval for the taxi data ? Could this have an impact on the refinements of the CDR based land-use characterization ? ------------------------------------------------------ ------------------------------------------------------ Reviewer B: Originality: How would you rate the originality of the paper?: Accept Relevance: How would you rate the relevance of this paper to the conference topics?: Accept Technical Quality: How would you rate the technical quality of this paper?: Weak Accept Presentation: How would you rate the presentation of this paper?: Accept Guidance for Authors: Please describe in detail main paper contributions, positive aspects, observed deficiencies, and suggestions on how to improve them:: The authors present an unsupervised learning technique to identify classes of signatures that are distinctive of different land use. They combined mobile phone data and GPS traces of floating vehicles provided by traffic operators. The presented results are very interesting and well-promising. All parameters of the developed technique are sufficiently described. The reader should definitely check the introduction paper that authors suggest in order to understand completely the proposed methodology. It is understandable that in a conference paper, whose extent is limited, it is not possible to fully develop the process. Datasets are described in detail. I noticed that for Milan and Naples the two datasets are from the same time period, whereas for Lyon they are not (GPS traces from 2012, mobile data from 2014). I would like to see a comment on that in the text. Why didn’t you choose the same time period as in the previous cases? The results presented in section V are interesting and Fig. 3 is also very useful in understanding the second level of classification. However, I would like to see how obvious the differentiation in at least one case is, by showing, for example, the OR (T1 Signature) AND the OR (T0 Signature).