公开(公告)号：US20160267381A1

公开(公告)日：2016-09-15

申请号：US15162505

申请日：2016-05-23

Applicant: salesforce.com, inc.

Inventor： Matthew Fuchs ,  Arun Jagota

IPC: G06N5/02 ,  G06N7/00 ,  G06F17/30

CPC classification number: G06N5/02 ,  G06F17/30292 ,  G06N7/005 ,  G06N99/005 ,  Y04S10/54

Abstract: The technology disclosed relates to methods for partitioning sets of features for a Bayesian classifier, finding a data partition that makes the classification process faster and more accurate, while discovering and taking into account feature dependence among sets of features in the data set. It relates to computing class entropy scores for a class label across all tuples that share the feature-subset and arranging the tuples in order of non-decreasing entropy scores for the class label, and constructing a data partition that offers the highest improvement in predictive accuracy for the data set. Also disclosed is a method for partitioning a complete set of records of features in a batch computation, computing increasing predictive power; and also relates to starting with singleton partitions, and using an iterative process to construct a data partition that offers the highest improvement in predictive accuracy for the data set.

Abstract translation: 所公开的技术涉及用于分配贝叶斯分类器的特征集合的方法，找到使分类处理更快更准确的数据分区，同时发现并考虑数据集中的特征集合之间的特征依赖性。 它涉及跨共享特征子集的所有元组的类标签的计算类熵分数，并且按照类标签的非递减熵分数的顺序排列元组，以及构建提供预测精度最高改进的数据分区 用于数据集。 还公开了一种用于分割批量计算中的特征记录的完整集合的方法，计算增加的​​预测能力; 并且还涉及从单例分区开始，并且使用迭代过程构造提供数据集的预测精度最高改进的数据分区。

公开(公告)号：US20160110354A1

公开(公告)日：2016-04-21

申请号：US14518145

申请日：2014-10-20

Applicant: salesforce.com, inc.

Inventor： Arun Jagota ,  Chenghung Ker ,  Parth Vijay Vaishnav ,  Danil Dvinov ,  David Hacker ,  Susan Levine

IPC: G06F17/30

CPC classification number: G06F17/30489

Abstract: Matching objects using keys based on match rules is described. A system generates a match rule key based on a match rule, wherein the match rule specifies whether two objects match. The system creates candidate keys by applying the match rule key to data objects. The system creates a probe key by applying the match rule key to a probe object. The system determines whether the probe key matches a candidate key. The system determines whether the probe object matches a candidate object based on applying the match rule to the probe object and the candidate object if the probe key matches the candidate key corresponding to the candidate object. The system identifies the probe object and the candidate object as matching based on the match rule if the probe object matches the candidate object.

Abstract translation: 描述使用基于匹配规则的键匹配对象。 系统基于匹配规则生成匹配规则密钥，其中匹配规则指定两个对象是否匹配。 系统通过将匹配规则键应用于数据对象来创建候选键。 系统通过将匹配规则键应用于探针对象来创建探测键。 系统确定探测键是否匹配候选键。 如果探测键与候选对象相对应的候选键匹配，则系统基于将匹配规则应用于探测对象和候选对象来确定探测对象是否匹配候选对象。 如果探测对象与候选对象匹配，系统将基于匹配规则将探测对象和候选对象标识为匹配。

公开(公告)号：US20150269647A1

公开(公告)日：2015-09-24

申请号：US14504593

申请日：2014-10-02

Applicant: salesforce.com, inc.

Inventor： Arun Jagota ,  Matthew Fuchs ,  Gregory Haardt

IPC: G06Q30/06

Abstract: A system creates a graph of nodes connected by arcs, and identifies a first compound attribute associated with contacts purchased by a current user. The first compound attribute includes a first attribute associated with a first value and a second attribute associated with a second value. The system identifies a directed arc from a first node to a second node. The directed arc is associated with a probability that previous users who purchased a first contact associated with the first compound attribute also purchased a second contact associated with a second compound attribute. The second compound attribute includes the first attribute, associated with a third value which matches the first value, and the second attribute, associated with a fourth value, which lacks a match with the second value. The system outputs a recommendation for the current user to purchase contacts associated with the second compound attribute if the probability exceeds a threshold.

Abstract translation: 系统创建通过弧连接的节点的图形，并且识别与由当前用户购买的联系人相关联的第一复合属性。 第一复合属性包括与第一值相关联的第一属性和与第二值相关联的第二属性。 系统识别从第一节点到第二节点的有向弧。 定向弧与购买与第一复合属性相关联的第一联系人的先前用户也购买了与第二复合属性相关联的第二联系人的概率相关联。 第二复合属性包括与第一值匹配的第三属性的第一属性和与第二值相关联的第二属性，其与第二值相匹配。 如果概率超过阈值，则系统输出当前用户购买与第二化合物属性相关联的联系人的建议。

公开(公告)号：US20150269595A1

公开(公告)日：2015-09-24

申请号：US14486111

申请日：2014-09-15

Applicant: salesforce.com, inc.

Inventor： Arun Jagota ,  Gregory Haardt ,  Govardana Sachithanandam Ramachandran ,  Lei Ming ,  Matthew Fuchs ,  George Vitchev ,  Fang Wong

IPC: G06Q30/02

CPC classification number: G06Q30/0202 ,  G06Q10/06315 ,  G06Q10/067 ,  G06Q10/103 ,  G06Q10/105 ,  G06Q30/00 ,  G06Q30/02 ,  G06Q30/0235 ,  G06Q30/0255 ,  G06Q30/06 ,  G06Q30/0631 ,  G06Q40/00

Abstract: Contact recommendations based on purchase history are described. A system creates a directed graph of nodes in which at least some of the nodes are connected by directed arcs, wherein a directed arc from a first node to a second node represents a conditional probability that previous users who purchased a first contact also purchased a second contact. The system identifies a set of contacts purchased by a current user. The system estimates a prospective purchase probability based on a historical probability that previous users purchased a specific contact and a related probability that previous users who purchased the specific contact also purchased a contact in the set of contacts, for each candidate contact. The system outputs a recommendation for the current user to purchase a recommended candidate contact based on a corresponding prospective purchase probability.

Abstract translation: 描述了基于购买历史的联系建议。 系统创建节点的有向图，其中至少一些节点通过定向弧连接，其中从第一节点到第二节点的有向弧表示先前用户购买第一个联系人的条件概率也购买了第二个节点 联系。 该系统识别当前用户购买的一组联系人。 该系统基于以前用户购买特定联系人的历史概率以及购买特定联系人的先前用户也为每个候选联系人购买联系人的联系人的相关概率来估计预期购买概率。 该系统基于相应的预期购买概率输出针对当前用户购买推荐候选联系人的建议。

公开(公告)号：US20140025683A1

公开(公告)日：2014-01-23

申请号：US13945758

申请日：2013-07-18

Applicant: Salesforce.com, Inc.

Inventor： Craig Howland ,  Stanislav Georgiev ,  Feng Meng ,  George Vitchev ,  Zandro Luis Gonzalez ,  Matthew Fuchs ,  Arun Jagota

IPC: G06F17/30

CPC classification number: G06F16/285 ,  G06F16/24578 ,  G06F16/335 ,  G06F16/9535 ,  G06Q10/107

Abstract: A method of managing crowdsourced data includes storing contact information regarding a plurality of contacts within a community-updateable repository accessible by a plurality of users, receiving a plurality of discrepancy reports associated with a selected contact of the plurality of contacts, extracting fact data regarding the selected contact from the plurality of discrepancy reports, determining an action to be taken based on the fact data and a fact model applied to the fact data, and performing the action to modify the community-updateable repository.

Abstract translation: 管理众包数据的方法包括存储关于多个用户可访问的社区可更新储存库内的多个联系人的联系人信息，接收与多个联系人的选定联系人相关联的多个差异报告，提取与多个联系人有关的事实数据 从所述多个差异报告中选择的联系人，基于事实数据确定要采取的动作，以及应用于事实数据的事实模型，以及执行修改社区可更新存储库的动作。

公开(公告)号：US20130117287A1

公开(公告)日：2013-05-09

申请号：US13667347

申请日：2012-11-02

Applicant: salesforce.com, inc.

Inventor： Arun Jagota ,  Pawan Nachnani

IPC: G06F17/30

CPC classification number: G06Q10/06

Abstract: A system and method for building a profile record for a person. Email addresses and corresponding person names are extracted from an email message and stored as records each record having an email address and corresponding person name as a key/value pair. A pair of such records is compared. If the person names are known for both records, then a match between the person names is evaluated. If the person name is known for only one of the records, then a match between the known person name for the one record and an email prefix for the other record is evaluated. If the person name is not known for either record, then a match between the email prefixes for both records is evaluated.

Abstract translation: 一种用于为人员构建简档记录的系统和方法。 从电子邮件消息中提取电子邮件地址和相应的人物名称，作为记录，将具有电子邮件地址和相应人物名称的每个记录作为密钥/值对存储。 比较一对这样的记录。 如果这两个记录的人名都是已知的，则评估人名之间的匹配。 如果人名仅为其中一条记录而知道，则评估该记录的已知人名与另一条记录的电子邮件前缀之间的匹配。 如果任何一个记录的人员名称都不知道，则两个记录的电子邮件前缀之间的匹配被评估。