基于倒排索引的DNF算法改进

中国互联网广告的高速发展使得广告市场对广告投放效果提出了更高的要求。布尔表达式检索作为定向广告的核心检索方式，决定着投放广告的精准度。由于具有倒排索引属性值唯一的特性，布尔表达式检索算法为广告主定制化需求带来了极大的限制。文中将使用流归并的思路改进DNF算法，在检索过程中归并同属性链表，从而实现用户标签属性的多值“与”关系索引建立及检索，更好地满足定向广告中广告主对定向条件多样化的需求。

Improved method of DNF algorithm based on inverted index

The rapid development of Chinese Internet advertising puts lorwaro mgn reqmrements totadvertising in the market. As the core retrieval mode of targeted advertising, DNF algorithm determinesthe accuracy of advertising. With the unique feature of inverted index, DNF algorithm sets greatrestrictions for customized demand of advertisers. This article improves the DNF algorithm using the ideaof stream merging, which can merge the linked list with the same attribute in the process, to realize themulti-value and relationship index for the attribute of the user tag, breaking the unique feature restrictionof the attribute in the DNF algorithm. Then the multi-value and relationship index for the attribute of theuser tag can be realized, advertisers can be better satisfied.