IR Search - what is your opinion

[kobear]

What is your opinion of the IR search and knowledge base that comes out of the box on SC 4.0.5 (4.x for that matter).



My experience is that the knowledge base is completely useless - just wondering if there are gotcha's that I am not considering, or is it just lacking.



For instance, in the knowledge base, you cannot search across all knowledge areas - you have to select individual areas. Also, the only area that returns any information to my queries (Incident Managemnent) returns records that have NOTHING TO DO WITH WHAT I AM SEARCHING FOR.



Frustrating.



I have developed my own knowledge base that is HTML based, and my management has asked me to justify why I did not use the internal knowledge base from service center, and I have a feeling they aren't going to be satisfied with "Because it just sucks". I want to know what everyone else has had experiences with this feature.



Thanks in advance

[tommy]

We are not on SC4 yet but I think the IR expert is basically the same in SC3 and SC4.



We have completely disabled the IR-expert for the following reasons:



- we run on mvs and we pay for cpu consumption and the IR expert is very costly in cpu time

- we had several examples of an IR search causing all updates to hang if a user had started a search that was taking a long time



But the users actually used the IR search and was more or less happy with the results.



Today we have a DB2 server that we update hourly with selected information, with that DB2 server we provide the helpdesk users a search functionality on our intranet.



Kyle, Do you have a good stopword file? It is essential to have a good stopword file for the IR expert to be efficient.



As for the knowledge database we really never tested that for real because 99% of all the solution the users marked for posible candidates for the knowledge database was solutions like:



"Done"

"Ok"

"Finished"



The knowledge database in SC5 has been improved a lot. This is described from memory - I attended a demo of SC5 today. It will be possible to categorize the knowledge on 7 new items:



- Category

- Subcategory

- Product type

- Problem type

- Asset

- Company

- and I can't remember the last thing.



This means that when searching for a solution when opening a call the first search takes into account all these fields including the free text field in the search. If this does not result in any hits you can remove the fields one by one and redo the search.



The results can be divided into 3 categories:

1. Hot News is always shown

2. Known Errors

3. Results based on the field search



It will be possible to add extra keywords on the knowledge records either manually or automatically. Also SC will learn which solutions is used most often and move them up in the list of hits.

[kobear]

Where is this stopword file located?

[d.schaefer]

Hi!



The stopword-file is located in the /data/-directory and it's named <ir_language>.stp

For example german.stp or english.stp.

It is a textfile and you can simply edit it with the notepad.

After editing don't forget to regenerate the indexes and restart ServiceCenter.



For further information have a look into the Data Administration Guide, Chapter 27 and 28.



Greetings from Germany

Daniel

[outlance]

- we had several examples of an IR search causing all updates to hang if a user had started a search that was taking a long time





Kyle, Do you have a good stopword file? It is essential to have a good stopword file for the IR expert to be efficient.

While trying to optimize a system with 110 users we had a deeper look into documentation about IR-Expert.

We found the following in Data Administration Guide (SC4)


ir_max_shared:nnnnn
This parameters sets the maximum amount of shared memory that you want
IR Expert to use. Versions 2 and later do not need to be completely in shared
memory, but rather uses the shared memory as a CACHE of the most
frequently referenced data. Increased shared memory improves performance.
...
...
The system determines if version 1.4 or a later version should be used by
checking the type field in the datadict entry for the file. If the type field
contains IR 2.0 then the IR Expert version 2 version is used. (The syntax must be IR in caps, a space and then 2.0.) If this data is not contained in the type field, the IR Expert 1.4 is used.


......

We just looked at the type field of all datadict with IR-Keys and all had:

type="IR"

So now we change it to "IR 2.0" and hope that it increases the performance.



A second result of our work was that we changed IR-Expert to
asynchronous mode (ir_asynchronous:1) and starting the
background process to read the irqueue file and write to IR-File.



I will post if we had success with it - or not!