TREX working for E-recruiting

[Former Member]

Hi all,

can anyone help me understand the following!

The standard system includes a range of periodic services that are required for the search within E-Recruiting. The search engine TREX (Text Retrieval and Information Extraction) uses search profiles and not operational data when searching. These are an extract of the operational data. To keep the data basis up to date for the search, the search profiles must be updated periodically. For the same reason, the documents that are included in the full-text search and the search profiles must be indexed from time to time and communicated to the search engine TREX. This indexing is performed by TREX and prepared and activated

with the INDEX_DOCUMENTS service.

- So indexing is done outside of TREX but done by TREX(how)!

what is operational data! How is it linked to TREX, indexing!

Regards,

Bharat

[Former Member · Accepted solution]

Hello,

Indexing is done by TREX, not outside of TREX.

Operational data is real data in database. Search profiles could be considered as
the extract of operational data. Their difference is, e.g. when you save a candidate
info into system, you can find it from operational data immediately after savng.
But you can find it from search profiles only after periodical service run(periodical
services pass the candidate data to TREX, then TREX indexes the data and puts the data
back into search profile afterwards).

Best Regards,
David

[Former Member · Accepted solution]

Thanks David for the reply.

I have some more queries:

1. Why should we use periodical services and do indexing while data is readily available in the database tables!

2. What are some common errors during indexing by TREX! how to deal with them!

3. As far as I know we would stop RCF_PERIODICAL_SERVICES when we do indexing. Is this right! If so why do we stop it! 

This is my view about the above point 3.

So first periodical services should run and send data to TREX.

Now since we have data with TREX we should not run periodical services because in case indexing starts at TREX there would be some incorrect data as periodical services would keep sending data even during indexing.

4. When we start search as recruiter or candidate we send request to database , which in turn would pass data to TREX. Now TREX does the indexing and would send it to the search profiles which would answer the search query. Right!

Please correct me if my understanding is wrong on any of the above!

Thanks,

Bharat

[Former Member · Accepted solution]

Hi Bharat,

1. The data already available in database are not indexed, so we need some services run to index those data. 

   System doesn't know when new data are inserted/updated into database, so we need some services run periodically to make sure the new data are also indexed.

2. I would suggest you look through below SAP notes:
   1400779 / 1345777 / 817145 / 815866

3. It is not right. You should keep it running as background job.
   There is one common scenario you need to stop RCF_PERIODICAL_SERVICES(for details, see note 1426757, "New attributes/information category fields").
   Note 1301016 describes another scenario.

4. The procedure you decribed is not correct.
   When you search(you should know not all the search are based on indexed data), indexed data are fetched from search profiles directly. It has nothing to do with TREX.

Hope this helps.

Regards,

David

[NicoleGeischnek]

Hello Bharat,

additional info to David's: Periodical job service INDEX_DOCUMENTS does the indexing of attachments, which you won't need anymore if you have activated business function HCM_ERC_SES_3 because then the attachments are stored automatically in the business object itself.

Regards

Nicole