Abstract Data encryption‐based and secret sharing‐based data outsourcing schemes protect the confidentiality of sensitive attributes but not their secrecy. Ciphertexts/shares generated by a data encryption/secret sharing scheme can attract the attention of interceptors. Thus, it is desired to hide the existence of highly‐sensitive attributes (as secret attributes) in the outsourced relations in addition to protecting their contents. This paper proposes a novel scheme that integrates data hiding with secret sharing for relational databases to protect both the secrecy and confidentiality of secret attributes. It embeds one or multiple secret attributes in a relation into one or multiple cover attributes in the same relation. A set of share (and possibly index) columns are constructed such that they are pretended to be associated with only the cover attributes, while those share columns and some virtual share columns can be used to recover both the secret and cover attributes. What interceptors observe in each relation include the attributes stored in plaintext and the share (and possibly index) columns associated with the cover attributes but not any extra column. Thus, they find nothing suspicious. This is the first effective data hiding scheme for relational databases that protects the secrecy of secret attributes.

Abstract Recommender systems nowadays play an important role in providing helpful information for users, especially in ecommerce applications. Many of the proposed models use rating histories of the users in order to predict unknown ratings. Recently, users’ reviews as a valuable source of knowledge have attracted the attention of researchers in this field and a new category denoted as review-based recommender systems has emerged. In this study, we make use of the information included in user reviews as well as available rating scores to develop a review-based rating prediction system. The proposed scheme attempts to handle the uncertainty problem of the rating histories, by fuzzifying the given ratings. Another advantage of the proposed system is the use of a word embedding representation model for textual reviews, instead of using traditional models such as binary bag of words and TFIDF 1 vector space. It also makes use of the helpfulness voting scores, in order to prune data and achieve better results. The effectiveness of the rating prediction scheme as well as the final recommender system was evaluated against the Amazon dataset. Experimental results revealed that the proposed recommender system outperforms its counterparts and can be used as a suitable tool in ecommerce environments.

Abstract Developers use software information sites such as Stack Overflow to get and give information on various subjects. These sites allow developers to label content with tags as a short description. Tags, then, are used to describe, categorize and search the posted content. However, tags might be noisy, and postings may become poorly categorized since people tag a posting based on their knowledge of its content and other existing tags. To keep the content well organized, tag recommendation systems can help users by suggesting appropriate tags for their posted content. In this paper, we propose a tag recommendation scheme that uses the textual content of already tagged postings to recommend suitable tags for newly posted content. Our approach combines multi-label classification and textual similarity techniques to improve the performance of tag recommendation. We evaluate the performance of the proposed scheme on 11 software information sites from the Stack Exchange network. The results show a significant improvement over TagCombine, TagMulRec and FastTagRec, which are well-known tag recommendation systems. On average, the proposed model outperforms TagCombine, TagMulRec and FastTagRec by 26.2, 15.9 and 13.8% in terms of Recall@5 and by 16.9, 12.4 and 9.4% in terms of Recall@10, respectively.

. Word Sense Disambiguation, which is one of the most challenging problems in the process of machine translation, can be considered as a classification problem. In this paper, we use K-Nearest-Neighbor, as one of the most popular classification methods, as well as some knowledge based resources in order to design a WSD scheme. The success of K-Nearest-Neighbor is tightly dependent on two factors; the features used to represent the context in which an ambiguous word occurs and the distance/similarity measure used for comparison of text vectors. Hence, in the present study, we focus on these two matters. For the first purpose, we extract three sets of features; syntactic features, lexical features and semantic features. In order to produce enriched and useful corpora, we apply preprocessed steps. In this work, we carry out a feature selection process as well as a feature weighting policy in order to fine-tune the classifier. For the second purpose, we try several distance/similarity metrics (rather than one metric) in order to find the most proper one. We also assign and use feature weights and propose a weighted formula for every metric. Moreover, to show that the proposed schemes are not language-dependent, we apply the suggested schemes to two sets of data; English and Persian corpora. The evaluation results, with regards to the feature selection and feature weighting strategies, show that the semantic and syntactic features have a significant effect on the classification ability of the system. The results are also encouraging compared to state of the art.

Abstract One of the major issues in the process of machine translation is the problem of choosing the proper translation for a multi‐sense word referred to as word sense disambiguation (WSD). Two commonly used approaches to this problem are statistical and example‐based methods. In statistical methods, ambiguity resolution is mostly carried out by making use of some statistics extracted from previously translated documents or dual corpora of source and target languages. Example‐based methods follow a similar approach as they also make use of bilingual corpora. However, they perform the task of matching at run‐time (i.e. online matching). In this paper, by looking at the WSD problem from a different viewpoint, we propose a system, which consists of two main parts. The first part includes a data mining algorithm, which runs offline and extracts some useful knowledge about the co‐occurrences of the words. In this algorithm, each sentence is imagined as a transaction in Market Basket Data Analysis problem, and the words included in a sentence play the role of purchased items. The second part of the system is an expert system whose knowledge base consists of the set of association rules generated by the first part. Moreover, in order to deduce the correct senses of the words, we introduce an efficient algorithm based on forward chaining in order to be used in the inference engine of the proposed expert system. The encouraging performance of the system in terms of precision and recall as well as its efficiency will be analysed and discussed through a set of experiments.

Frequent itemset mining over sliding window is an interesting problem and has a large number of applications. Sliding window is a widely used model for frequent itemset mining over data streams due to its capability to handle concept drift, its bounded memory and its low processing time. A sliding window-based algorithm requires an efficient data structure that can be updated as fast as possible when inserting and deleting transactions. Moreover, an innovative computing method is needed to provide the set of frequent patterns (FPs) with a little delay after the user issues a request for the mining results within a window. In this study, an efficient representation of the sliding window named blocked bit sequence is introduced which is aimed to store and maintain the content of the window. Moreover, by a novel technique this representation is exploited for efficiently extracting the set of FPs within the current window. Experimental evaluations on both real-life and synthetic data streams show that the proposed approach is faster than recently proposed algorithms in different phases of data stream mining.

Word sense disambiguation (WSD) can be thought of as the most challenging task in the process of machine translation. Various supervised and unsupervised learning methods have already been proposed for this purpose. In this paper, we propose a new efficient fuzzy classification system in order to b e applied for WSD. In order to optimize the generalization accuracy, we use rule-weight as a simple mechanism to tune the classifier and propose a new learning method to iteratively adjust the weight of fuzzy rules. Through computer simulations on TWA data as a standard corpus, the proposed scheme shows a uniformly good behavior and achieves results which are comparable or better than other classification systems, proposed in the past.