Numerous logs generated during the daily deployment and operation process in cloud computing platforms help system administrators perform anomaly detection. Common anomalies in cloud workflow include pathway anomalies and time delay anomalies. Traditional anomaly detection methods train the learning models corresponding to the two kinds of anomaly detection tasks respectively and ignore the correlation between these two tasks, which leads to the decline of the accuracy of anomaly detection. In order to solve the problems, an anomaly detection method based on multi-task temporal convolutional network was proposed. Firstly, the event sequence and time sequence were generated based on the event templates of log stream. Then, the deep learning model based on the multi-task temporal convolutional network was trained. In the model, the event and the time characteristics were learnt in parallel from the normal system execution processes by sharing the shallow layers of the temporal convolutional network. Finally, the anomalies in the cloud computing workflow were analyzed, and the related anomaly detection logic was designed. Experimental results on the OpenStack dataset demonstrate that, the proposed method improves the anomaly detection accuracy at least by 7.7 percentage points compared to the state-of-art log anomaly detection algorithm DeepLog and the method based on Principal Component Analysis (PCA).

Because the existing Web quality assessment approaches rely on trained models, and users' interactions not only cannot meet the requirements of online response, but also can not capture the semantics of Web content, a data Quality Assessment based on Simulated Annealing (QASA) method was proposed. Firstly, the relevant space of the target article was constructed by collecting topic-relevant articles on the Web. Then, the scheme of open information extraction was employed to extract Web articles' facts. Secondly, Simulated Annealing (SA) was employed to construct the dimension baselines of two most important quality dimensions, namely accuracy and completeness. Finally, the data quality dimensions were quantified by comparing the facts of target article with those of the dimension baselines. The experimental results show that QASA can find the near-optimal solutions within the time window while achieving comparable or even 10 percent higher accuracy with regard to the related works. The QASA method can precisely grasp data quality in real-time, which caters for the online identification of high-quality Web articles.

As existing inverted index not only cannot efficiently handle multiple-keyword query, but also cannot find results for misspelled keywords, a bi-level index leveraging Bitmap and Locality-sensitive Hashing (BLH) was proposed to support multiple-keyword queries. The upper-level of BLH is bitmaps, which map keywords onto clusters of sub-graphs based on the n-grams in the keywords. Each cluster stores the similar sub-graphs. On the lower-level, each cluster has a locality-sensitive hashing index, which helps identify the sub-graphs that contain the keywords based on their n-grams. The indexing scheme of BLH can dramatically decrease query I/Os, thus reducing the query time by 80%. Furthermore, the index based on n-gram can avoid the sensitivity to spelling mistakes of keywords, guaranteeing to return expected results in any case. The experimental results on real data sets demonstrate the effectiveness of the BLH index, which can efficiently support the querying of Web and social network.

The existence of potential links in sparse networks becomes a big challenge for link prediction. The paper introduced active learning into the link prediction task in order to mine the potential information of a large number of unconnected node pairs in networks. The most uncertain ones of the unlabeled examples to the system were selected and then labeled by the users. These examples would give the system a higher information gain. The experimental results in a real communication network dataset Nodobo show that the proposed method using active learning improves the accuracy of predicting potential contacts for communication users.

Focusing on the applications of wireless sensor networks in marine monitoring, a localization algorithm of sensor nodes based on Monte Carlo was proposed. The algorithm improved the location accuracy of the prediction stage by introducing the right angle of the sea motion; and revised the prediction coordinates by the confidence of the prediction coordinates which was gained through the nodes perception of pressure. The simulation results show the proposed algorithm has outstanding performance compared with traditional methods under conditions of different node density, seed density, moving speed and time.