如今,互联网集成的与暴雨灾害相关的信息多种多样,然而人工搜索网页信息的效率不高,因此网络主题爬虫显得十分重要。在通用网络爬虫的基础上,为提高主题相关度的计算精度并预防主题漂移,通过对链接锚文本主题相关度、链接所在网页的主题相关度、链接指向网页PR值和该网页主题相关度的综合计算,提出了基于网页内容和链接结构相结合的超链接综合优先度评估方法。同时,针对搜索过程易陷入局部最优的不足,首次设计了结合爬虫记忆历史主机信息和模拟退火的网络主题爬虫算法。以暴雨灾害为主题进行爬虫实验的结果表明,在爬取相同网页数的情况下,相比于广度优先搜索策略(Breadth First Search,BFS)和最佳优先搜索策略(Optimal Priority Search,OPS),所提出的算法能抓取到更多与主题相关的网页,爬虫算法的准确率得到明显提升。
Recently, Li et al. presented a two-party quantum private comparison scheme using Greenberger-- Horne-Zeitinger (GHZ) states and error-correcting code (ECC) lint. J. Theor. Phys. 52, 2818 (2013)], claiming it is fault-tolerant and could be performed in a non-ideal scenario. However, there ex- ists a fatal loophole in their private comparison scheme under a special attack, namely the twice- Hadamard-CNOT attack. Specifically, a malicious party may intercept the other party's particles and execute Hadamard operations on the intercepted particles as well as on his or her own particles. Then, the malicious party could sequentially perform a controlled-NOT (CNOT) operation between intercepted particles and the auxiliary particles, as well as between his or her own particles and the auxiliary particles prepared in advance. By measuring the auxiliary particles, the secret input will be revealed to the malicious party without being detected. For resisting this special attack, a feasible improved scheme is proposed by introducing a permutation operator before the third party (TP) sends the particle sequences to each participant.