Released 2024.11.12

• Previously, when a non-existent matching was sent to the v1/inbtrck/externalMatching endpoint, a 500 error was returned. We rectified this behavior. Now, the API will return the more accurate 422 error code.

Released 2024.10.15

• New feature: You can now natively integrate Generative AI in Inbenta Search. A new /generate-answer endpoint is available to obtain an answer using Generative AI. You can find the relevant documentation in the Developers' Portal.

Released 2024.10.01

• Fixed: When a request is made to POST /federated-search or POST /federated-searches, in cases when the Search API cannot retrieve some contents, those contents will be discarded from the Inbenta Semantic Search Engine results. This avoids errors and returns retrievable contents as normal.

Released 2024.08.06

• As part of a common effort to improve the security and performance of Inbenta products, we have upgraded the technology behind the Search API.

Released 2024.07.09

• The Search API now allows sending the x-inbenta-compress header. If this header is set to 1, API responses will be returned compressed in gzip format.

Released 2024.04.16

• Fixed: An error occurred when spell corrections were applied to a query and the instance had an attribute that could be interpreted as a number. This is now fixed.

Released 2024.04.03

• We added the possibility of tracking matchings with contents that do not belong to the instance's knowledge base, but to an external knowledge base.

Released 2024.01.23

• As part of a common effort to improve the security and performance of Inbenta products, we have upgraded the technology behind the Search API.

Released 2023.05.03

• Added: The API routes for event tracking were sometimes blocked by browser ad blocker extension. This was mainly because they contained the word “tracking“ as part of the path. To avoid this, we modified the tracking routes names, replacing “tracking” with “inbtrck”. The old endpoints will continue working, but we highly recommend using the new ones.

Released 2023.03.07

• We added the new parameter “context” to the payload of the POST /tracking/events and the POST /federated-search endpoints of the Search API, which will be registered in the “User Information” column of the table in Dashboard > Sessions > Details.

Released 2023.02.07

• Fixed: Some validation errors occurred in the POST /federated-searches endpoint, but not in the POST /federated-search endpoint. We made the validation consistent in both endpoints and also improved some error messages for better understanding.

• Added: We removed the “sessionId” parameter from the POST /tracking/session endpoint.

Released 2022.12.13

• Fixed: Some errors occurred when registering content events in Search+ instances. This is now fixed.
• Fixed: When the Search API received a /tracking/events request without specifying a contentId, an internal error was reported. This is now fixed.
• Added: We made a small improvement to optimize the performance of the Search API when retrieving answers.

Released 2022.10.18

• We updated the POST /tracking/events endpoint to allow customers to track their conversion events to calculate their own conversion rates. This was a requirement for the development of a new conversion events report in Search instances.

Released 2022.08.09

• Fixed: The Search endpoints were not returning all results when they received an empty query with the splitQuery parameter set to true. This is now fixed.

Released 2022.06.28

• Fixed: Queries sent to the federated-searches endpoint are no longer tracked multiple times. If the same query is sent in two different requests, it will only be tracked once.
• Added: We added validation errors for filters and sorting criteria. In case a filter or a sorting criterion is specified (via the federated-search and federated-searches endpoints) and the attribute used is not marked for use as a filter or for sorting, an error is returned.

Released 2022.06.14

• Fixed: Dynamic abstracts generated for each result were not generated considering the selected user type. This means that the dynamic abstract may have displayed information from an unintended user type. It is fixed now, and it only displays the content attribute value of the requested user type.

Released 2021.10.19

• Added: We added score as one of the options available in our API to sort contents. It allows sorting contents by the matching score returned for a given query. It can help to create custom sorting logics when different results have the same score. Find more information here.

Released 2021.04.20

• Added: To increase the security in our system, we removed the Set-cookie header returned from the API.

Released 2021.04.06

• Fixed: The Search API returned a warning when the access token had expired. It is fixed.

Released 2021.03.23

• Fixed: Search API must split the user query internally when it is longer than 255 characters. This was not working for all cases, causing delays in processing some user queries.

Released 2020.07.22

• Fixed: A bug was found where, when someone used the Search API and set the parameter type as null in the /federated-search and /federated-searches endpoints, the API returned a 500 error. This is now fixed.

Released 2020.04.15

• Added: We created a new endpoint called "/language-detector" in the Search API. This endpoint helps users identify a different language in a query. For more information, see Search API Routes. 

Released 2020.04.01

• Updated: We updated Inbenta rate-limits.

Released 2020.02.19

• Added: We made several improvements to the Search API to provide new filtering types (e.g. greater than, less than, etc.). For more information, see Search API Routes.

Released 2019.11.13

• Added: We adapted the API to track Knowledge Base events and clicks on contents received without performing a query. For more information, see API Tracking Guidelines.

Released 2019.10.16

• Fixed: there was a recurrent error returned internally by the Search API: Specified ‘contentId’ doesn’t exist. This error most likely returned a 500 error. This is now under control.

Released 2019.10.02

• We removed a workaround that made the API handle an engine issue related with Thai instances with Dynamic abstracts highlighting (See Release 2019.09.04). Since the issue was fixed, the workaround was no longer needed.

Released 2019.09.18

• From now on, the Search API endpoints /federated-search and /federated-searches responses will provide two new parameters for each result found: creationDate and modificationDate. These parameters represent the date when the content was first indexed and last modified, respectively.

Released 2019.09.04

• Fixed: A recurrent error was reported when the response of the API contained a Dynamic Abstract in a language with different characters, such as Thai. This is now fixed. 

Released 2019.08.07

• New: We added a new Advanced Search Option to Federated-search/Federated-searches, that allows you to set the max number of characters that a query should have to split. The option is called ‘applySplitQueryMaxCharacters’ and can be found in the ‘advancedSearchOptions’ property.

Released 2019.07.24

• You can now use quotation marks to set exact words (e.g. How to achieve "objectives")  or expressions (e.g. "Trial master file" ) in a search query. When you do this, the search will return this exact term (not synonyms) or group of terms in the same order. This improvement allows Search users to be much more precise in their queries.

Released 2019.07.10

• The API now returns all failed filters when in Debug mode, to show in the Search Debug tool.

Released 2019.05.15

• New: The Search API can now track information about the SDK version used as part of the User information.
• Fixed: A bug was found where under some circumstances the API was not returning a highlightedTitle: it returned null, and its fallback value was to return the title without highlighting. This is now fixed.

Released 2019.04.17

• New: There is a new sorting logic by Popularity. This new sorting sorts results by the amount of clicks these contents received over time. 
• New: There is a new set of tools in the API and SDK to allow the UI to filter by empty attributes. This enhances the full functionality, and allows you to create new tabs like "Other results", or to add contents without filtering value into the filtering logic.

Released 2019.04.03

• The /tracking/session endpoint of the API now returns not only the sessionToken, but also the sessionId. This allows developers to retrieve and use sessionId to filter results in Dashboards. 

Released 2019.03.20

• The Search API was always returning a 401 status code for any ICS issue. This is fixed and it now returns the proper status code.
• We optimized the API modules to improve performance in production.

Released 2019.03.06

• The API was not retrieving the Alternative titles when they where matching the user question. This is now fixed.

Released 2019.02.06

• After splitting a user question into different queries, the API does a merge of the different values to eliminate duplicates. While the deduplication process was "random", now it favours the contents with higher scores.
• Reviewed all of the API modules to enhance the performance of the search endpoints

Released 2019.01.23

• Fixed: Due to the internal cache of the API, some content attributes were not returned in the results if the query has already been asked during the same session.

Released 2019.01.09

• Fixed: When using the maxResults parameter on /federated-search or /federated-searches endpoints to limit the number of results, if the splitQuery was set to true, the API could return more results than maxResults. This is no longer the case. 
• If x-inbenta-env in API requests is set as "development", the request will point to fromsite "Internal test" in the Search App. 

Released 2018.12.19

• New: Added new maxResults parameter on the /federated-search and /federated-searches endpoints. This lets users modify the maximum number of results on their searches, which improves performance.
• Fixed: The API sometimes returned a 500 error. This error was caused by the codification of certain characters in certain languages when the API tried to split the query. This is now resolved.

Released 2018.12.12

• Added a new way to rotate the signing keys that generate the access tokens. This allow Inbenta to rotate the signing key without impacting clients.
• When using the splitQuery parameter, the API will only process the first non cached query it finds. This is done to prevent errors processing large chunks of text that are usually copy-pasted on deflection tools
• Updated rate limits library

Released 2018.11.21

• Fixed: A bug was found where sorting was not executed properly. This is now fixed.

Released 2018.11.07

• Fixed: Force sorting if there is more than one query.
• [Hotfix] Update JOSE library

Released 2018.10.17

• Added: Updated API and SDK tracking documentation.
• Fixed: Added session logs for contact_start, contact_submit and contact_ticket events.

Released 2010.10.03

• Group the same User Questions under the same LogId by Session for /federated-searches endpoints
• Updated rate-limits
• Fixed: Removed non functional Slack webhook
• Fixed: /federated-search in instant answers takes from 2 to 4 seconds

Released 2018.09.19

• Added: Group the same User Questions under the same LogId by Session for /federated-search and /federated-searches endpoints
• Fixed: Solved vulnerability found in user types

Released 2018.09.05

• Fixed: Filters containing ":" were not parsing correctly. They do now.

Released 2018.08.22

• Added: There is now a limit to the number of subrequests in /federated-searches

Released 2018.08.08

• Updated: The Logs flow was updated to:
  • Add session to deflection tools
  • Auto-log searches in session.

Released 2018.07.25

• Added:  there is a new endpoint at /app/data/{dataId}
• Improvements have been made to the filtering algorithm.
• API can now return groups of attributes
• Fixed: the Content Digest crawler should not allow URLs or blank spaces as titles

Released 2018.06.13

• Log filtering information as User information
• Use different kinds of highlighting

Released 2018.05.30

• Fixed: Sort by relevance should respect ICS order

Released 2018.05.16

• Add highlighting to titles
• Send cache headers for the autocompleter endpoints
• Return original attributes that matched a query

Released 2018.05.02

• Added: Ability to obtain all contents doing an empty query
• Removed creation of log entries for internal filtering actions

Released 2018.04.18

• Fixed: Increase memory limits to allow more compiled data