基于BIDE的多核并行闭合序列模式挖掘

基于经典的BIDE算法,提出一种多核并行闭合序列模式挖掘算法——MT_BIDE。该算法在频繁序列扩展判断前进行剪枝,在扩展过程中动态调整频繁序列及其伪投影数据集,平衡不同线程间挖掘闭合序列模式的计算量差异。实验结果表明,该算法具有较高的运行效率和加速比。

A comparative study of changing patterns of concanavalin A-binding proteins in early stage of cholesterol gallstone

IM To elucidate the importance and the changing patterns of biliary concanavalin Abinding proteins (CPs) in the early stage of cholesterol gallstone.METHODS CPs concentrations and nucleation activities were measured by lectin affinity chromatography in biles of patients with cholesterol gallstone, pigment gallstone, gallbladder cholesterosis and nonbiliary diseases.RESULTS The concentrations of CPs were much higher in patients with cholesterol gallstone (039g/L±011g/L, n=36, P<001) or gallbladder cholesterosis (040g/L±009g/L, n=9, P<001) than in those with pigment gallstone (026g/L±012g/L, n=7) and/or nonbiliary diseases (027g/L±009g/L, n=10). Pronucleating activities were much stronger in patients with cholesterol gallstones (nucleation time ratio: 057±021, n=5, P<001 vs pigment gallstone and/or nonbiliary diseases) and gallbladder cholesterosis (nucleation time ratio: 044±023, n=5, P<001 vs pigment gallstone or nonbilliary diseases). The binding percentages of CPs to model biliary vesicles were also higher in patients with cholesterol gallstones (n=6) than those with pigment gallstones (n=6) (24%±09% vs 09%±05%, P<001).CONCLUSION Hypersecretion of CPs, especially those in vesicular phase may be the important changes in the early stage of cholesterol gallstone.

一种基于用户行为模式的匿名数据鉴定方法

针对匿名用户数据的海量性与冗余性等特点,为提高数字证据的用户身份鉴定性能,文章提出基于用户行为模式的匿名数据鉴定方法。首先,文章研究了基于BIDE算法的用户频繁行为模式挖掘方法,为数据鉴定提供了高质量的用户频繁序列行为模式库。然后,采用基于最长公共子序列的相似度方法得到模式综合相似度,全面描述用户数据之间的吻合程度。最后,分别使用Web浏览数据集和Unix操作命令行数据集进行实验,结果表明,文章所提出的数字证据鉴定方法具有良好的适用性和计算效率,为匿名数据的同一鉴定提供了技术支撑。