Workday, Inc.

Google Inc.

Parag Avinash Namjoshi

Harikrishna Narayanan

Vishwanath Tumkur Ramarao

Mark Risher

Naveen Jamal

Cory Mitchell O'Connor

Sonal Rajan

Tejaswi Nadahalli

Youngnam Maeng

Mohammad Sabah

Hamdi Jenzri

Yury Markovsky

Vladimir Giverts

Sergei Winitzki

Adam Charles Baker

Lakshminarayanan Renganarayana

Shuangshuang Jiang

Daniel Walter Beck

Kristy Gateley

Adeyemi Ajao

Pavan Boob

Xiao Fan

Michael Au

Manjunath Balasubramaniam

James Fan

Kevin Mun Joun Tham

A system for validating data includes an interface and a processor. The interface is configured to receive a data set. The processor is configured to calculate a data quality metric for the data set, wherein the data quality metric is based at least in part on a data distribution metric; determine a model to build based at least in part on the data quality metric; build the model; and provide the model for use.

Data validation for automatic model building and release

A system for improving a query response includes an interface and a processor. The interface is configured to receive a query. The processor is configured to determine a categorization for the query using a model. The categorization is associated with a confidence value. The processor is configured to a comparison of the confidence value and a first threshold, wherein the first threshold is determined to maximize a metric and modify the query response based at least in part on the categorization and the comparison.

Model validation for query intent

A secure tenant activity data system for a deployment service includes processor that is configured to decrypt a set of encrypted stream processed monitoring data to recover a set of stream processed monitoring data; determine a transaction identifier associated with the set of stream processed monitoring data; determine whether the transaction identifier has been seen associated with a set of previous data; in response to determining that the transaction identifier has been seen associated with the set of previous data, retrieve the set of previous data; decrypt the set of previous data to recover a set of decrypted previous data; aggregate the set of stream processed monitoring data with the set of decrypted previous data to generate a set of aggregated data; encrypt the set of aggregated data to create a set of encrypted aggregated data; and provide the set of encrypted aggregated data to a safe storage area.

User action collection for sensitive customer data

A secure tenant activity data system for a deployment service includes an interface and a processor. The interface is configured to receive a set of encrypted monitoring data collected and encrypted from a monitoring instrument from a tenant system. The processor is configured to decrypt the data for stream processing; encrypt and store the stream processed data; and provide the encrypted data with a tenant identifier to a deployment service. Steam processing is one or more of data cleaning, data rounding, data filtering, data smoothing, algorithmic processing, improving ranking of results, improved navigation, generating recommendations, generating metrics, or computation of features for a machine learning system.

A system for improving a query response includes an interface and a processor. The interface is configured to receive a query. The processor is configured to determine a categorization for the query using a model. The categorization is associated with a confidence value. The processor is configured to determine whether the confidence value is above a first threshold and, in response to the confidence value being above the first threshold, modify the query response based at least in part on the categorization.

A system for generating a database of labeled foreign canonical titles includes an interface and a processor. The interface is to receive a title in a second language. The processor is to 1) store a set of n-grams in a first language in a first database; 2) sanitize the title into a sanitize title in the second language; 3) translate the sanitized title into a translated title in the first language; 4) break the translated title into n-grams; 5) determine labels for the n-grams using the first database; and 6) determine label to associate with the title.

Systematic mass normalization of international titles

A system for determining retention risk comprises a grouper, a filter, a normalizer, a feature vector extractor, a model builder, and a predictor. The grouper is for determining a set of time series of transactions where each is associated with one employee. The filter is for filtering the set of time series of transactions based on an employee transition characteristic to determine a subset of time series. The normalizer is for determining a model set of time series by normalizing the subset of time series. The feature vector extractor is for determining a set of feature vectors determined from a time series of the model set of time series. The model builder is for determining one or more models based at least in part on the set of feature vectors. The predictor is for predicting retention risk for a given employee using the one or more models.

Retention risk determiner

A system for identifying address components includes an interface and a processor. The interface is to receive an address for parsing. The processor is to determine a matching model of a set of models based at least in part on a matching probability for each model for a tokenized address, which is based on the address for parsing, and associate each component of the tokenized address with an identifier based at least in part on the matching model, wherein each component of the set of components is associated with an identifier, and wherein probabilities of each component of the set of components are determined using training addresses.

Address parsing system

A system for identifying address components includes a training address interface, a training address probability processor, a parsing address interface, and a processor. The training address interface is to receive training addresses. The training addresses are a set of components with corresponding identifiers. The training address probability processor is to determine probabilities of each component of the training addresses being associated with each identifier. The parsing address interface to receive an address for parsing. The processor is to determine a matching model of a set of models based at least in part on a matching probability for each model for a tokenized address, which is based on the address for parsing, and associate each component of the tokenized address with an identifier based at least in part on the matching model.

The problem of abuse of privileges at cloud-hosted service sites is addressed by connecting a single individual or 'actor' with multiple user accounts and/or other online identities, thereby creating a 'consolidated profile.' In this way a confidence level can be established that a particular user account, IP address or other identifying attribute is associated with a particular actor. Different confidence levels may suffice depending on the remediary action to be taken; for example, holding a message for human review is obviously less draconian than rejecting the actor's registration at sign-up, and would therefore require a lower degree of confidence.

Detection and prevention of unwanted content on cloud-hosted services

The problem of abuse of privileges at cloud-hosted service sites is addressed by associating each user, preferably across multiple cloud-hosted service sites, with an individual 'suspiciousness' score that may vary over time as additional user actions are detected and evaluated. Knowledge of the user is employed to better analyze the appropriateness or acceptability of user actions on the site.

The problem of abuse of privileges at cloud-hosted service sites is addressed at the sign-up stage by identifying suspicious or abusive users and preventing them from signing up in the first place. This approach may utilize a relatively small initial data set based on the recognition that while abusers' profiles are mutable and difficult to characterize stably&#x2014;abusers will deliberately shift their usage patterns to evade detection&#x2014;a site's legitimate users tend to have similar and stable characteristics; that is, such characteristics are 'cohesive' across cloud-hosted service (e.g., social media) sites and their users. The information gleaned from analysis of this small data set can be applied to a much larger, unsorted data set to obtain profiling criteria based on a large population for statistical reliability.

Growing community of inventors

Foster City, CA, United States of America

Parag Avinash Namjoshi