公开(公告)号：US20100125585A1

公开(公告)日：2010-05-20

申请号：US12272607

申请日：2008-11-17

申请人： Wei Chu ,  Seung-Taek Park ,  Raghu Ramakrishnan ,  Bee-Chung Chen ,  Deepak K. Agarwal ,  Pradheep Elango ,  Scott Roy ,  Todd Beaupre

发明人： Wei Chu ,  Seung-Taek Park ,  Raghu Ramakrishnan ,  Bee-Chung Chen ,  Deepak K. Agarwal ,  Pradheep Elango ,  Scott Roy ,  Todd Beaupre

IPC分类号： G06F17/30 ,  G06F7/06 ,  G06F7/00

CPC分类号： G06F16/3346 ,  G06F16/313

摘要： Information with respect to users, items, and interactions between the users and items is collected. Each user is associated with a set of user features. Each item is associated with a set of item features. An expected score function is defined for each user-item pair, which represents an expected score a user assigns an item. An objective represents the difference between the expected score and the actual score a user assigns an item. The expected score function and the objective function share at least one common variable. The objective function is minimized to find best fit for some of the at least one common variable. Subsequently, the expected score function is used to calculate expected scores for individual users or clusters of users with respect to a set of items that have not received actual scores from the users. The set of items are ranked based on their expected scores.

摘要翻译： 收集关于用户，项目以及用户和项目之间的交互的信息。 每个用户与一组用户特征相关联。 每个项目与一组项目特征相关联。 为每个用户 - 物品对定义预期分数函数，其表示用户分配项目的预期分数。 目标表示用户分配项目的预期分数与实际分数之间的差异。 预期得分函数和目标函数共享至少一个共同变量。 目标函数被最小化以找到最适合至少一个共同变量中的一些。 随后，使用预期分数函数来计算相对于尚未从用户那里获得实际分数的一组项目的个体用户或用户群的预期分数。 该组项目根据其预期分数进行排名。

公开(公告)号：US20100030717A1

公开(公告)日：2010-02-04

申请号：US12184114

申请日：2008-07-31

申请人： Deepak Agarwal ,  Pradheep Elango ,  Raghu Ramakrishnan ,  Seung-Taek Park ,  Bee-Chung Chen

发明人： Deepak Agarwal ,  Pradheep Elango ,  Raghu Ramakrishnan ,  Seung-Taek Park ,  Bee-Chung Chen

IPC分类号： G06N5/02

CPC分类号： G06Q30/02

摘要： Content display policies are evaluated using two kinds of methods. In the first kind of method, using information, collected in a “controlled” manner about user characteristics and content characteristics, truth models are generated. A simulator replays users' visits to the portal web page and simulates their interactions with content items on the page based on the truth models. Various metrics are used to compare different content item-selecting algorithms. In the second kind of method, no explicit truth models are built. Events from the controlled serving scheme are replayed in part or whole; content item-selection algorithms learn using the observed user activities. Metrics that measure the overall predictive error are used to compare different content-item selection algorithms. The data collected in a controlled fashion plays a key role in both the methods.

摘要翻译： 使用两种方法评估内容显示策略。 在第一种方法中，使用以“受控”的方式收集关于用户特征和内容特征的信息，生成真实模型。 模拟器会根据真实模型重播用户对门户网页的访问，并模拟与页面上的内容项目的交互。 各种指标用于比较不同的内容项目选择算法。 在第二种方法中，没有建立明确的真理模型。 受控服务计划的活动部分或全部重播; 内容项目选择算法学习使用观察到的用户活动。 衡量总体预测误差的度量用于比较不同的内容项目选择算法。 以受控方式收集的数据在这两种方法中起关键作用。

公开(公告)号：US06681222B2

公开(公告)日：2004-01-20

申请号：US09906502

申请日：2001-07-16

申请人： Navin Kabra ,  Raghu Ramakrishnan ,  Uri Shaft ,  Vuk Ercegovac

发明人： Navin Kabra ,  Raghu Ramakrishnan ,  Uri Shaft ,  Vuk Ercegovac

IPC分类号： G06F1730

CPC分类号： G06F17/30675 ,  G06F17/3053 ,  Y10S707/99932 ,  Y10S707/99933 ,  Y10S707/99934 ,  Y10S707/99935

摘要： A unified database/text retrieval system converts exact database type queries into text inclusion type queries suitable for text retrieval systems through the use of pseudo keywords. Boolean combination of the text inclusion type query elements may be readily manipulated for optimization and applied to a unified index for rapid search results. Absolute relevance values and relevance multiplier values may be added to the query elements to provide a relevance-based sorting not only of text but also of exact match type search results. Relevance values may be deduced automatically from a variety of sources.

摘要翻译： 统一的数据库/文本检索系统通过使用伪关键字将精确的数据库类型查询转换为适合文本检索系统的文本包含类型查询。 文本包含类型查询元素的布尔组合可以容易地被操纵以用于优化并应用于用于快速搜索结果的统一索引。 可以将绝对相关性值和相关性乘数值添加到查询元素中，以提供不仅文本的相关性排序，而且还提供精确匹配类型搜索结果的基于关联的排序。 相关性值可以从各种来源自动推导出来。

公开(公告)号：US06442561B1

公开(公告)日：2002-08-27

申请号：US09465203

申请日：1999-12-15

申请人： Johannes E. Gehrke ,  Venkatesh Ganti ,  Raghu Ramakrishnan

发明人： Johannes E. Gehrke ,  Venkatesh Ganti ,  Raghu Ramakrishnan

IPC分类号： G06F1730

CPC分类号： G06F17/30705 ,  G06F2216/03 ,  Y10S707/99943

摘要： A method of creating and updating a binary decision tree from training databases that cannot be fit in high speed solid state memory is provided in which a subset of the training database which can fit into high speed memory is used to create a statistically good estimate of the binary decision tree desired. This statistically good estimate is used to review the entire training database in as little as one sequential scan to collect statistics necessary to verify the accuracy of the binary decision tree and to refine the binary decision tree to be identical to that which would be obtained by a full analysis of the training database.

摘要翻译： 提供了一种从不适合高速固态存储器的训练数据库创建和更新二进制决策树的方法，其中可以适应高速存储器的训练数据库的子集用于创建统计上良好的估计 需要二进制决策树。 这种统计学上的良好估计用于以少至一个顺序扫描来检查整个训练数据库，以收集必要的统计信息，以验证二进制决策树的准确性，并将二进制决策树细化为与通过 全面分析培训数据库。