Recommender Service security and privacy

Here you can find an overview of the Jive Genius Recommender Service security and privacy.

If you have enabled the Recommender Service (it is enabled by default), your Jive instance periodically collects information about users' behavior and send it to the service hosted in Jive Software's data center. This service is delivered from a secure data center in the United States. The Recommender Service relies on state-of-the-art security technology, including a multi-stage encryption process that uses a powerful AES-256 encryption algorithm, to ensure your data is secure and private. Your instance locally encrypts your data into tokens using your unique customer key. This key is available only on your instance. Therefore, it is impossible for the Recommender Service to decrypt any of your data. The Recommender Service processes only data from keyword counts, content types, and user and content relationships.

Before any data is submitted to the Recommender Service, all textual data (subject and content) is processed for efficiency and security. First, keywords for the content in your instance are calculated locally and then encrypted with a unique key stored locally and known only to your instance. Then, your local instance collects activity data about how users have interacted with content, people, and places within your community. The activity data and keywords are then further encrypted and transmitted to the Recommender Service over a secure SSL channel.

The Recommender Service determines that multiple people are interested in the same encrypted keyword without knowing the true value of the keyword. The Recommender never processes the actual textual content from your instance; it processes encrypted data. In addition, the data sent to the service cannot be reversed in the Recommender. This allows the Recommender to provide social intelligence about what users find interesting without requiring access to your sensitive business data. Results are then returned to your instance, decrypted locally as necessary, and delivered to your community users.