7

The use case:

We are creating a system which analyses mail content. If we are able to categorize and archive an email based on its content, we want to automatically categorize subsequent emails in the same conversation / thread.

The issue is that the conversationId we get as part of the (mail)message is not unique when we query on behalf of different users.

An example:

A: The conversationId for a given thread in my inbox.

B: The conversationId for the same thread, but in a different user's inbox.

Values:

A:AAQkADUyZWYxNzljLTc4NjItNGMzYy1iZDYwLTE4NWEzNDg1OWUzZQAQABlanHOPI0v1ukA7KePaYv4=. B:AAQkAGVkNGRjNWNmLTAwNTItNDA2NC1hOThhLTU5NTUyNGFjNTM5ZAAQABlanHOPI0v1ukA7KePaYv4=.

Can I use parts of this conversationId to identify the same conversation across different users? The pattern seems to be that the last 25-26 characters are unique for the given conversation. The rest is a Guid + a postfix of some sort.

Edit: My question is related to usage of the Microsoft Graph API. There are questions answered on SO related to ConversationIds and Outlook emails, but these describe cases getting emails direct from the outlook client (or maybe through the outlook api). For instance are properties like the ConversationIndex and ConversationTopic not in the message I get from the Graph API.

Namig Ismayilov
  • 121
  • 17
Stig Nielsen
  • 71
  • 4
  • Possible duplicate of [How to get universal outlook mailitem ConversationID for one email chain replied from different users](https://stackoverflow.com/questions/14823129/how-to-get-universal-outlook-mailitem-conversationid-for-one-email-chain-replied) – Namig Ismayilov May 11 '19 at 02:55
  • 1
    Thank you for finding this one @NamigIsmayilov, but the message described in that thread has different properties compared to the ones I get from the Graph API, I am still hoping that parts of the ConversationId have some uniqueness across multiple users. – Stig Nielsen May 24 '19 at 10:17
  • Did you find a resolution for this? – David Clarke Oct 02 '19 at 03:06
  • @DavidClarke, sorry, no proper resolution for this one. I ended up using the assumption that the last 25 characters are unique, so these are what I use to identify conversations. This could easily break if Microsoft decides to change anything their side, so not sure I can recommend it. – Stig Nielsen Jan 31 '20 at 09:05

0 Answers0