Release Notes - March 7th 2017


#1

A couple of releases for LinkedIn Engagement Insights are pending. Please find details about the upcoming changes below.

Rate Limit Changes

We’re changing the way rate limiting work on LinkedIn Engagement Insights to allow you to queue a larger number of tasks for processing.

Each user will have a queue of 1,000 queries that will be processed in accordance with the user’s rate limit. Please note that rate limit are applied hourly from the point in time of the first query no matter when this is submitted. Rate limits are applied per ‘master’ account and are shared by any identities under that account.

Two new headers are added to the POST /pylon/{service}/task response object which will tell you the size of your queue every time you submit a query.

  • X-Tasks-Queued
  • X-Tasks-Queue-Limit

Results are stored for 7 days before being deleted.

Example

Let’s assume a user wants to queue up 1,001 queries. The user can queue up the first 1,000 and the last query gets rejected. As soon as one (1) query has been processed (and therefore removed from the queue) the user can queue up the last query. As such, a user can at any time have up 1,000 queries queued up. New queries can be added to the queue as long as there are <1,000 queries in the queue.

Case sensitivity for query filters

To avoid confusion around case sensitivity we will apply the following rules for targets with string values.

  1. The ==, != and IN operators are case sensitive.
  2. The contains, contains_any, contains_near, contains_all operators are case insensitive.

These rules will apply to all targets of type ‘string’ and ‘array(string)’. Any targets with these types added in future will keep to these rules.

New type system for entities called ‘Concepts’

PYLON for LinkedIn Engagement Insights applies named entity recognition to articles, and you can use these to better understand what drives engagement and interest amongst audiences. You can also filter data by entities; making it possible to craft new audiences around interests and focuses.

There are usually many entities in play across any articles. LinkedIn Engagement Insights will apply up 20 per article. To help put structure to these we are introducing ‘concepts’ as a new namespace that will replace the entities namespace.

Please note: we will retire li.all.entities on 15th April 2017.

Concepts are the same as entities, but now includes the name and type of the entity. The types are shown below:

  • City
  • Company
  • Country
  • Education
  • Entertainer
  • Government Agency
  • Location
  • Misc
  • Organization
  • Person
  • Political Party
  • Politician
  • Product
  • Sportsperson

The new concepts namespace contains three fields that you can use for filtering and analysis - all fields are tokenized:

  1. li.all.concepts.names: The entities detected for articles similar to li.all.entities.
  2. li.all.concepts.types: The types for entities attached in li.all.concepts.names.
  3. li.all.concepts.type_names: A pipe separated field of type and names.

Upcoming changes to LinkedIn Engagement Insights :: Concepts