The ability to explain in understandable terms, why a machine learning model makes a certain prediction is becoming immensely important, as it ensures trust and transparency in the decision process of the model. Complex models, such as ensemble or deep learning models, are hard to interpret. Various methods have been proposed that deal with this matter. Shapley values provide accurate explanations, as they assign each feature an importance value for a particular prediction. However, the exponential complexity of their calculation is dealt efficiently only in decision tree-based models. Another method is surrogate models, which emulate a black-box model's behavior and provide explanations effortlessly, since they are constructed to be interpretable. Surrogate models are model-agnostic, but they produce only approximate explanations, which cannot always be trusted. We propose a method that combines these two approaches, so that we can take advantage of the model-agnostic part of the surrogate models, as well as the explanatory power of the Shapley values. We introduce a new metric, Top j Similarity, that measures the similitude of two given explanations, produced by Shapley values, in order to evaluate our work. Finally, we recommend ways on how this method could be improved further.

The topic of this paper is the exploration of the various characteristics of the wavelet tree data structure, a data structure that was initially proposed for text compression applications but has found a plethora of other uses in text indexing and retrieval. Issues concerning the efficient maintenance of the structure, plus its handling in various applications are explored. Our main aim is to provide to computer science researchers that would like to explore the specific area, an up-to-date comprehensive material covering a wide range of applications. This kind of up-to-date survey is missing from the current bibliography and we hope that it will help young researchers to get familiar with the notions of this research area.

The identification of influential users in social media communities has been recently of major concern, since these users can contribute to viral marketing campaigns. In our approach we extend the notion of influence from users to networks and consider personality as a key characteristic for identifying influential networks. We describe the Twitter Personality based Influential Communities Extraction (T-PICE) system that creates the best influential communities in a Twitter network graph considering users' personality. We then expand existing approaches in users' personality extraction by aggregating data that represent several aspects of user behavior using machine learning techniques. We use an existing modularity based community detection algorithm and we extend it by inserting a pre-processing step that eliminates graph edges based on users' personality. The effectiveness of our approach is demonstrated by sampling the twitter graph and comparing the influence of the created communities with and without considering the personality factor. We define several metrics to count the influence of communities. Our results show that the T-PICE system creates the most influential communities.