In this paper, we study the impact of data complexity and data quality in the overlapping community detection problem. We show that community detection algorithms are very unstable against incomplete or erroneous data, and this result is consistent with all the evaluated performance metrics. We verify it using three quality metrics (F1, NMI, and Omega) when the ground-truth community structure is known, in four very popular and representative detection algorithms: Order Statistics Local Optimization Method (OSLOM), Greedy Clique Expansion (GCE) algorithm, Speaker-listener Label Propagation Algorithm (SLPA), and Cluster Affiliation Model for Big Networks (BIG-CLAM). We evaluate it over a set of real instances that arise from detecting the courses that belong to different careers (degrees) of an engineering University, and over large benchmark sets of synthetic instances frequently used in the literature.

Word embeddings are widely used in natural language processing (NLP) to group semantically similar words but have been applied in other areas to find semantic similarity between entities. In this paper we create a vector embedding for Internet Domain Names (DNS) using a corpus of real anonymized DNS log queries from a large Internet Service Provider (ISP). We then use this embedding as a layer of a recurrent neural network (RNN) that works as a Language Model for the DNS queries generated by the users. We show that this RNN can be used to predict the next DNS query generated by a user with good accuracy (considering the size of the problem). Moreover, we show that training the same RNN without using the pre-trained vector model takes more time and is substantially less accurate. The results presented in this work can have practical applications in many engineering activities related to DNS architecture design. For example, latency reduction in address resolution, optimization of cache systems in recursive DNS servers, automatic filtering of inappropriate domains, and detecting anomalies in traffic.

Word embeddings is a well known set of techniques widely used in natural language processing (NLP), and word2vec is a computationally-efficient predictive model to learn such embeddings. This paper explores the use of word embeddings in a new scenario. We create a vector representation of Internet Domain Names (DNS) by taking the core ideas from NLP techniques and applying them to real anonymized DNS log queries from a large Internet Service Provider (ISP). Our main objective is to find semantically similar domains only using information of DNS queries without any other previous knowledge about the content of those domains. We use the word2vec unsupervised learning algorithm with a Skip-Gram model to create the embeddings. And we validate the quality of our results by expert visual inspection of similarities, and by comparing them with a third party source, namely, similar sites service offered by Alexa Internet, Inc.

Going to the cinema or watching television are social activities that generally take place in groups. In these cases, a recommender system for ephemeral groups of users is more suitable than (well-studied) recommender systems for individuals. In this paper we present a recommendation system for groups of users that go to the cinema. The system uses the Slope One algorithm for computing individual predictions and the Multiplicative Utilitarian Strategy as a model to recommend to an entire group. We show how we solved all practical aspects of the system; including its architecture and a mobile application for the service, the lack of user data (ramp-up and cold-start problems), the scaling fit of the group model strategy, and other improvements in order to reduce the response time. Finally, we validate the performance of the system with a set of experiments with 57 ephemeral groups.