Mining User-aware Rare Sequential Topic Monitoring Pattern on Single Sign-on Document Stream in Multiple Web Applications

In this paper, Sequential Topic Patterns (STPs) technique is used to formulate the issues of User-aware Rare Sequential Topic Patterns (URSTPs) mining in Internet document soure. The Sequential Subject Pattern (STP) is used to define and track Internet users' customised and abnormal behaviours. In certain real - world contexts, STP is incorporated, such as tracking of irregular user behaviours. A set of algorithms are used in three stages to overcome innovative mining issues: first, pre-processing to retrieve probabilistic topics and define sessions for various users. Second, using pattern-growth, generating all the STP candidates with (predicted) support factors for each user. Third, by doing user-aware rarity evaluation on derived STPs, choosing URSTPs.

Collaborative Access to Ancient Documents

With the evolution of the next generation networks several applications have emerged to be used through the web. Applications allowing the analysis and the recognition of documents are emerged to be used through Internet. Document pre-processing output may affect the efficiency of document analysis and recognition systems. In order to ameliorate the efficiency of such systems, an objective evaluation of pre-processing steps is necessary. The authors propose a new framework for automatic evaluation of binarization approaches. The evaluation of binarization approaches is based on the comparison between binary images and their ground-truths. For that reason a method for the generation of ground-truth is proposed. This method is evaluated using the benchmarking dataset DIBCO 2009. The comparison between binarization methods’ results is based on several evaluation metrics as FM and PSNR. The proposed framework is tested on printed images (250 images) from the Google-Books collection and on handwritten images (60 images) from the IAM historical database.