Example-Based Taxonomy

Hi Experts,
Could someone of you give the example document to configure example-based taxonomy.
What exactly mean by example document. Does it has specific format...???
Thanks,
Satya

Hi Satyadev,
refer to these links:
<a href="http://help.sap.com/saphelp_nw04s/helpdata/en/bf/a17a1dd15211d6b2c700508b6b8a93/content.htm">What is an Example-based Taxonomy</a>
<a href="http://help.sap.com/saphelp_nw04s/helpdata/en/6c/5145b1d1de11d6b2cc00508b6b8b11/content.htm">Creating a taxonomy</a>
<a href="http://help.sap.com/saphelp_nw04s/helpdata/en/6c/5145d3d1de11d6b2cc00508b6b8b11/content.htm">Creating an example-based taxonomy</a>
<a href="http://help.sap.com/saphelp_nw04s/helpdata/en/08/19d83d4d855610e10000000a114084/content.htm">Creating an example-based taxonomy from scratch</a>
<a href="http://help.sap.com/saphelp_nw04s/helpdata/en/9d/19d83d4d855610e10000000a114084/content.htm">Creating an example-based taxonomy using a folder hierarchy</a>
<a href="http://help.sap.com/saphelp_nw04s/helpdata/en/a7/833c3cf0344e4de10000000a11402f/content.htm">Taxonomy Trainer</a>
Best regards,
Denis

Similar Messages

Cannot train an example based taxonomy in trex

Hi friends,
I have created an index in index administration.And attached a data source also.Then i create a new example based taxonomy and also i create folders for categories.Then in content management i go to taxonomy trainner.There is no choice of upload of document through which i can train the taxonomy.Is there a way through which i can upload the documents in the categories i have created.Also the training documents should be in what format(doc,ppt).Also if somebody have a document for Example based and query based taxonomy creation please send the link.
I am having EP6 sp15 and trex 6.1.
Thanks

Hi
In example or query based taxonomy you have to upload the doccuments in the KM repository that is the data source whic is a folder in KM repository framework right click and then use upload the docs into it and in taxonomy you only create categories and manually train the taxonomy using query builder or trainer in Classification inbox .I will suggest you to go through the guides available in SERVICE market place service.sap.com and use HOW-To search term and then navigate to TREX to find the guides.Docs can be any format.
You can use an existing hierarchy in your CM to train your new taxonomy.
You can train a taxonomy automatically if you already have a folder hierarchy that you want to transport to a taxonomy.
Create the taxonomy in index management. This determines that you are going to train the taxonomy using an existing folder hierarchy. The system carries out the following steps:
It copies the specified folder and its entire subordinate folder hierarchy to a taxonomy structure. This creates a category for each folder.
The system then carries out the initial training of the taxonomy.
All documents in the folder hierarchy are also indexed. You can decide whether the folder that you are using to train the taxonomy is to be assigned to the index.
You train the taxonomies initially by classifying documents from your folder hierarchy.
This initial training is necessary so that the system can carry out automatic classification, or can make suggestions when manual classification takes place.
For the initial training, you classify documents from the folders in your knowledge management system. You do this is the iView in which you navigate within the folder hierarchy.
The necessary number of documents depends on the quality of the documents
hope this helpe you please do not forget to give points
with regards
subrato kundu

Query-based taxonomy doesn't work

Hi,
I created a query-based taxonomy in the index of a folder. Then in Content Management - Classification I set the Category and query of the taxonomy. However, documents that meet the query cannot be classified into that category - they are still in Documents to Classify. It's the same with example-based taxonomy.
Am I missing something?
Thanks,
Ray

Hi Ray,
To work with this kind of taxonomy you must do:
1. Create index using classification service.
2. Create taxonomy based in query.
3. If you click in this taxonomy you can create new folders where documents will be stored (as result of classification, logical links).
4. In Content Management->Classification->Taxonomy Query Builder->choose the taxonomy created in step 2 and
create a search query for a category in a query-based taxonomy.
http://help.sap.com/saphelp_nw2004s/helpdata/en/77/49523ec744d947e10000000a114084/frameset.htm
Patricio.

Example based classification not working

Hi,
I am trying to set up example based classification.
I have done the below steps
1) Created an index
2) Assigned data sources
3) Created Taxonomy and categories
4) Trained "test.txt" with word "test" inside to a specific taxonomy category as defined above.
When i upload another file which is a copy of "test.txt" using Content Admin --> KM Content, it is not classifying automatically. I have done a re-index after the same also.
I am not sure which part i am missing out?
I have also set up QBT which works fine.
Appreciate your inputs.
Regards
PN

After a lot of training on sample documents the classification worked

Query Based Taxonomy and Searching Problem!

Hi all,
Ive an urgent problem to solve ....
I create some customs properties with dependent values, for example, if I put Country: Germany, I ve three options, (Waldorf,Munchen,Berlin) and so on ...
Then, I create a Query Based Taxonomy, where in some categories Ive a query, with the dependent value (Ex: Waldorf) ... I mean, If City = Waldorf, the document belongs to this category.
So, when I do this, the taxonomy fails ... Only with the dependent value, I mean, if the category is with the base property (Ex, Country) it works.
Plese, help!
Cheers,
Andres.

Hi,
To include a dependant property in a query, you have to enter both parent (Country = Germany) and the dependant (City = Munich) properties in the category.
Hope this help
Kind regards
Vincent

Query based taxonomy from scratch gives NoSuchMethodError...

I'm trying to create a query based taxonomy following the instuctions specified in the help docs: KM Platform -> Admin Guide -> Content Mgmt -> Taxonomies and Classification -> Creating a QBT -> from scratch. I created an index on a newly created (hence empty) folder in the 'documents' repository. So, essentially, the data source points to that folder. Also, I selected 'TREX Classification' on the 'Service' drop-down while creating the index. I left the crwler profile as blank. Then I went on to create a new 'Query Based Taxonomy'. I was able to create it alright, but as soon as I click on it to add folder and the such, I get this error:
java.lang.NoSuchMethodError
     at com.sapportals.wcm.repository.manager.taxonomy.TaxonomyNamespaceManager.getDocumentTaxRMRids(TaxonomyNamespaceManager.java:1876)
     at com.sapportals.wcm.repository.manager.taxonomy.TaxonomyNamespaceManager.getTaxonomyClassResources(TaxonomyNamespaceManager.java:1831)
     at com.sapportals.wcm.repository.manager.taxonomy.TaxonomyNamespaceManager.addChildDocuments(TaxonomyNamespaceManager.java:1942)
     at com.sapportals.wcm.repository.manager.taxonomy.TaxonomyNamespaceManager.getChildren(TaxonomyNamespaceManager.java:368)
     at com.sapportals.wcm.repository.CollectionImpl.internalGetChildren(CollectionImpl.java:966)
     at com.sapportals.wcm.repository.CollectionImpl.getChildren(CollectionImpl.java:179)
Any help in solving this is highly apprecaited.
We run EP6 SP2.
I did NOT create a 'crawler profile' for this set up.
thanks,
Biju.

Hi Biju,
re. your NW'04 SP stack 4 migration intention:
NW'04 is still in ramp-up (SP stack 4 is the current shipment bundle of it) so at the current point in time, only participating ramp-up customers can use it.
A migration from EP6.0 SP2 to NW'04 will be possible (but not during the ramp-up phase).
From a Knowledge Management & Collaboration perspective, we will port most of the NW'04 SP stack 4 functions to EP6.0 SP2 as well. These functions will be shipped as EP6.0 SP2 Patch 5 (planned shipment date beginning of Q4 / 2004) so for mid-term, you can also plan to stay on EP6.0 SP2 if you have no other pressing reasons for going to NW'04 (e.g. enhanced Web AS functions).
This feature porting from NW -> EP6.0 SP2 is an 'only-once' action, however, and NetWeaver will be the release for new features in the future.
Regards,
Joerg

Error while creating Query based taxonomy.

Hi,
I have created an index and trying to build a query based taxonomy though
Taxonomy Query Builder in content management. Earlier I have created
another index but after that I have deleted it but still while trying to build a query based
taxonomy it is giving an error that earlier index doesn't exist and not diaplaying the new
index and taxonomy that I have created. Taxonomy Trainer is working fine.
Please suggest to resolve this problem.
Thanks,
Deepika

Hi,
did u created the taxonomy in the new index are you imported.
Generally once you delete an index and you try to create an index with the same name wil give error. This may be because of the cache problem. Try to restart the server. Hope that may help you.
Raghu

Query Based Taxonomy - Dynamic Creation of a Query Problem

Hi,
I'm trying to create new category nodes in a taxonomy and to assing them queries. I know how to do this, but I encountered a strange behavior. Here is the situation:
I have a string property DocumentType. When manualy creating a query, it looks like:
DocumentType CONTAINS "Some string"
But when I do the same using code I get:
DocumentType IS "Some string"
And as a result of this, the taxanomy is not working properly.
Here is the code:
     indexService =
     (IIndexService) ResourceFactory
     .getInstance()
     .getServiceFactory()
     .getService(
     IServiceTypesConst.INDEX_SERVICE);
     //        This kind of classification index offers functionality working with query based taxonomies
     qbcIndex = (IQBClassificationIndex) indexService.getIndex(indexId);
     //        Get a query folder - a query folder is a representation of a query based taxonomies category
     queryFolder = qbcIndex.getQueryFolder(RID.getRID(folder));
     //        List of IQueryEntry objects for the query folder
     queryEntryList = queryFolder.getQuery(context);
     //        Cast the IQBClassificationIndex into a ISearchIndex object
     searchIndex = (ISearchIndex) qbcIndex;
     queryEntry = searchIndex.getNewQueryEntry();
     queryEntry.setRowType(IQueryEntry.ROW_TYPE_ATTRIBUTE);
     queryEntry.setPropertyName(propertyName);
     queryEntry.setPropertyOperator(IQueryEntry.PROPERTY_OPERATOR_EQUAL);
     queryEntry.setValue(propertyValue);
     queryEntry.setTermAction(IQueryEntry.TERM_ACTION_EXACT);
     queryEntryList.add(queryEntry);
     //        Set the query for the query folder
     queryFolder.setQuery(queryEntryList, context);
     qbcIndex.updateCategoryContents("Document_Types", false, context);
Can somebody tell me what am I doing wrong please. Isn't it because of IQueryEntry.PROPERTY_OPERATOR_EQUAL? Is there any substring operator?
Thanks a lot.

Hi, Robert
In this moment, I have a problem with Query Based Taxonomy, maybe could you help me. I insert the query in to the taxonomy from code, but when i use the method
qbcIndex.updateCategoryContents("/taxonomies/borrame5/Prueba2", false, context);
I got the next exception:
#1.5#0014C210F3D1002A000001200000565B000431CC93AD5979#1180654627543#System.err#sap.com/irj#System.err#dummy01#10241#####SAPEngine_Application_Thread[impl:3]_17##0#0#Error##Plain###     at com.sap.engine.core.thread.impl3.SingleThread.run(SingleThread.java:170)#
#1.5#0014C210F3D100310000012B0000565B000431CCC47166AB#1180655445698#System.err#sap.com/irj#System.err#dummy01#10241#####SAPEngine_Application_Thread[impl:3]_18##0#0#Error##Plain###com.sapportals.wcm.service.indexmanagement.IndexNotReadyException: Some documents are not included in this update because indexing has not been finished yet.#
#1.5#0014C210F3D100310000012C0000565B000431CCC4717417#1180655445701#System.err#sap.com/irj#System.err#dummy01#10241#####SAPEngine_Application_Thread[impl:3]_18##0#0#Error##Plain###     at com.sapportals.wcm.service.indexmanagement.retrieval.trex.AbstractTrexIndex.updateCategoryContents(AbstractTrexIndex.java:2493)#
I have doubt what parameters i need to pass to the method updateCategoryContents,
Let me give some information:
taxonomy:
   /taxonomies/borrame5/Prueba2
repository:
   /documents/Portal_KOF/KM/Pruebas1/Pruebas Versionamiento
classification and search index:
   borrame5
   I can't classify automatically, when i insert the query from code. When I manually insert the query from the portal, automatic classification run perfectly.
Thanks
Greetings
JARO.

Documents in Query-based Taxonomy are not Autoindexing

Hello experts!
I am on EP6.14 with Trex 6. I have a repository with a query-based taxonomy on it and is also part of a regular search index. I have autoclassification on.
On each document I have a couple of properties that are used to match documents in queries in my taxonomy. On the initial set-up and load, everything worked great. Documents went right where they were supposed to. But sometime after we released this to the business, the documents no longer automatically index. If I change a document property or set up a new one, it finds the category but leaves them as "suggested." You then have to go in and accept those suggestions to get the document to the right taxonomy. In short, it acts as if autoclassification is off. I have noticed that the Classification index does not submit a new document if I change a property but the search index does. Not sure if that is applicable but I would think both should do that since it is a change.
I did some research and found that synchronization issues or new folders could cause this problem but I have synchronized with no success. Also, state management is not on for the repository.
Is there any reason why autoclassification would stop on its own? What can I do to get it back on?
Thanks-
J
Many points given for helpful answers!

Hi Frederick-
Thanks for your response-
We have a single CM DB repository. There is a search only index at the root folder and classification index on a folder below the root. Normally, we don't overlap but we wanted only the folders in the lower folder to be part of the taxonomy. I suppose we could just rule out the other documents by not having them match any queries but this way it was guaranteed.
THe search and classifications are working (aren't throwing errors) but like I said, I would think that both would have to push the document through the queue. But I'm only seeing the search index do that. Now if I go and accept the recommended categories, then it goes into the queue for the Classification index.
J

Any extension examples based on JSR 198

I am looking for any extension samples available for JDeveloper EA based on JSR-198.
Are there any exiting? Will be helpful if you can point to some (even a beta version of the examples are ok)

I am looking for any extension samples available for JDeveloper EA based on JSR-198.
Are there any exiting? Will be helpful if you can point to some (even a beta version of the examples are ok)

Creating folder Structure for Query based taxonomy

Hello friends,
I am confused as to where do I create a folder hierarchy for taxonomy. Do I need to create a new repository for creating a folder structure. The help.sap.com mentions about using the Category Browser iview for creating the same. I am unable to find this iview in EP6 SP12. Can someone, please guide me as to where do I create a taxonomy folder structure hierarchy?
Your help is appreciated.
Thanks,
G.G

Hey G.G.,
Yes, i'd love to.
So, as stated above, directly below the "Taxonomies" folder you'll see the names of the existing taxonomies. There is no "new" option on this level, because a new taxonomy can only be created under "System Administration" -> "System Configuration" -> "Knowledge Management" -> "Index Administration". After you create a "TREX Search and Classification" or "TREX Classification" index, you can go to the index menu "Taxonomies" and say "New". This will automatically create a new folder under in the "Taxonomies" view.
Now, if you go one level further, so second subfolder under "Taxonomies" you should have the "New" option. If not, then you'll have to check the permission settings. Please go to "Content Administration" -> "KM Content" -> "Taxonomies" and then on the Menu "Details" -> "Settings" -> "Permissions". You might need to adjust permissions as suggested by SAP in SAPNote (599425):
Role:"ContentManager"=FULL CONTROL;
Group:"Everyone"=READ.
Please check that the already existing subfolders have inherited the permission changes.
Hope this solves your problem,
Robert

Taxonomy - how many training documents needed for automatic classification?

Greetings,
I have an example based taxonomy.
I have trained the taxonomy manually with about two documents and expected it to classify automatically after that which it has not done.
According to the URL below
http://www.scribd.com/doc/7234683/Retrieval-and-Classification-TREX-Features-and-Functions
it is mentioned that "You can also classify documents automatically. A prerequisite for this is that there are existing document classes with sufficiently correctly classified documents that meet the classification criteria"
Please advise if training the taxonomy with only two documents which I did was too little - exactly how many documents are needed? (the above URL does not give an indication of a number)
Regards,
Keith

Still trying to solve the problem, adding and tailoring documents as best as I can.

Taxonomies question

Hello,
We are using example-based taxonomy with automatic classification (KMC 6.0.16). I like to know that when I delete a taxonomy (from the Index Administration iView), does the training documents done previously are lost ? If I re-create the same taxonomie, are they still available and are used as the basis for automatic classification.
Thank you in advance.

Hi Patricio,
Thank you for your reply. I also thought the same, that the training documents are lost when I delete the taxonomy from Index Administration iView. However, as soon as I re-create the same taxonomy, the trained documents are again used as basis for automatic classification. What do you think of this.
Thank you.

Tip on Using "Folder" for performing Query Based Classification

In setting up a TREX taxonomy for one of our intranet sites we concluded that the folder structure used by the web developers really helped classify the documents at a higher than an 80% level. It seems that in many cases, the intranet, outlook public folders, or LAN folders will be organized such that the title of the folder is very valuable in classifying what is in it.
After many searches and false starts, I found something that is so simple and elegant, I had to share it here.
In the Taxonomy Query Builder there is a property "Folder". Unlike the property "Content" which has a "contains" operator, this one is a "Is" or "Is Not" so it seemed like it might be hard to use.
What I found is that it automatically put an * at the end of whatever we entered and if we put an * in the beginning, it seemed to find the folder consistently.
So for a three level taxonomy such as:
-- Business
Customer Support
   Product information
where you have folders /Business/CustSupport/ProdInfo you use a three level taxonomy.
For the business I put in a Query:
-- Folder is_not PassTheIsNotTestSoEverythingIsClassified
That meant everything was included at the root level no matter what. (i.e. I didn't want 'documents to be classified' to contain anything.)
The query for the Customer Support level of the taxonomy was:
   -- Folder is *CustSupport
which pulls in any document that is in the folder that contains CustSupport
The query for the next level was:
   -- Folder is *ProdInfo
which pulls in any document that is in the folder that contains ProdInfo
As the classification engine works, items that pass all three test are put in the most detailed folder. You can have as many taxonomy nodes at any level as you need and if they don't pass any, they are kept at the root of last level they did pass.
You end up with an amazingly simple classification scheme that can handle a big taxonomy. To the extent your folders are not organized like your taxonomy or are not related logically, they you need to add in other queries using the document contents, or other fields available for indexing.   This is where the 80-20 rule kicks in, you still need to work hard to get that last 20%.
Let me know if this approach helps anyone.

HI
r u trying to define properties at the document level then it doesnot works because metadata of KM and metadata of document are different.
u can define pre-defined properties at the folder level in KM.
why dont you use example based classification as u train ur taxonomies with example and it inturn index furher document automatically based on your example document.
But it still deponds upon individual requirements.
Regards,
Vijay.

Problems with Taxonomy and Classification

We are currently on EP6 SP2 and are going live in a couple of weeks. Currently, we are stuck on our solution to one of the requirements and cannot seem to find a way out of it. I would really appreciate anyone's help in this regard.
We have a web repository that has HTML files with meta-tags in them. The meta-tags have been defined using the "keywords" tag e.g.
<META NAME="keywords" CONTENT="oranges, lemons, limes">
We also have an index that is linked to a query-based taxonomy. The queries have been defined to read the embedded-keywords field and match it to the content in the HTML tag e.g. oranges, lemons etc. in the above example.
When we run the index, we know that it has picked up all the right HTML because the results are returned in the Search. However, they never get classified. All documents just keep sitting in the "Documents to be Classified" folder.
There are two kinds of errors we have come across. For some reason, TREX is not finding the field and is returning this error:
[00208] 2005-06-10 18:19:24.324 e attributes AttributeStore.cpp(00220) : openAttribute() failed: 6915 UNKNOWN_ATTRIBUTE (AttributeEngine); indexId = "aa_ys_sitedirectoryindex760001en"; attribute = "embedded-keywords"
This is even though we see embedded-keywords as a property and as something that can be assigned a query in the Taxonomy Query Builder.
Additionally, if you go and look at the details of a document that we are trying to classify, it has the following error on it:
An error occurred during classificationError while classifying resource /mytaglb/sitmeta_2/mil.dla.loc.HQ.user/Adoption.html for taxonomy News_and_Information.com.sapportals.wcm.WcmException: taxonomy not found (Errorcode 2422)
Even though the Taxonomy is there.
We have no idea why this is happening and have spent the last week or so exclusively looking at this problem. If anyone has any insight, please let us know as this is at a critical level for us.
Thanks a lot in advance.

hi Yasser,
Even I posted a message which is a bit related to your problem.
Your problem is that you assumed the functionality of "embedded-keywords" and used it in your query and it is not working as you expected and is throwing an error saying UNKNOWN-ATTRIBUTE.
I donno the difference b/n "name" and "name(as string)" attributes. The problem is we are not able to find specific help topic relating to the attribute options we have here. Can anybody tell us the location of proper documentation relating to these attributes found in the Query Builder?
For the message raised by me I got valuable reply from "Matthias Roebig-Landau". Hope he might help you too, if he has experience with the attribute "embedded-keywords".
Else, since your requirement is very urgent, better you can raise an OSS message to SAP.
Best regds,
Alagammai.

Example-Based Taxonomy

Similar Messages

Maybe you are looking for