CIS categorize without populating attribute

I have a taxonomy node "Access Controls" which contains several levels of access control values, for our purposes just imagine "Classified, Secret, Top-Secret", etc. I would like for documents to be categorized into this node based on a object attribute value but I don't wan't any attributes populated based on what is found only categorization based on the object's attribute value. Other nodes will need to do full text entity extraction with categorization and classification, only this node will not. I could create a class and a second attribute just to store the values populated by CIS but I'd rather not do that.

Any ideas?

Find more posts tagged with

BPM

Comments

Cyril

Hi,

I read two different subjects in this question: categorizing on attribute values vs. on full-text, and storing assigned category names in a document attribute. I'm answering on the second subject, don't hesistate to continue with another question if it is not what you need.

With 2 different taxonomies A and B, it is not possible to have CIS store assigned category name in document classification attribute if category belongs to taxonomy A, and not if it belongs to B.

I suggest you contact support with a detailled use-case to have a change request created so it is considered in a future release.

Thanks,

Cyril

fainemr

Thanks, Cyril, it does sound like two subjects and I guess they are related.

I have one existing taxonomy with several classes, so far this has been configured for full-text based categorization and storing category names in attributes and I would like this to continue to be the case except I would also like for one set of categories (and its subcategories) to only categorize based on an existing attribute value while the rest of the taxonomy continues to categorize based on full-text and stores category names in an attribute.

The root of the problem is that CIS cannot determine this value from the full text and so we will need to rely on it being set manually for the time being, but we would still like for these documents to be categorized using this attribute value.

Cyril

Hi,

CIS provides 2 ways to categorize based on existing attribute values: category property rules, and docset (or default) attribute processing configuration.

Property rules in category let you express simple constraints on documents attribute values. When no keyword rules (aka "evidence") is provided for a category, then a verified property rule triggers document categorization. The categories under "Access Controls" should contain property rules, instead of evidence terms.

Content processing configuration let CIS consider attribute values as content for category keyword rules. The attribute values can replace full-text content, or be added to it. Adding the attribute content to full-text might work in your case, but if the attribute values are too common in the full-text then it would generate too many false categorization.

These two approaches are documented in CIS Administration Guide (page 63 and 83 in version 6.7). Does one solve the issue?

Regards,

Cyril

fainemr

Yes, I'm aware of how it works and yes I'm afraid that content processing would result in too many false categorizations.

I do intend on using property rules in the categories under "Access Controls", however, these categories still must have a category class and if they have a category class this class will have a classification attribute, and this classification attribute will then be populated with the categorziation results. This is the problem.

I wan't the categorization but I do not want any attributes (in these categories) to be populated based on the categorization result. To be clear, only categorization should be the result for the category property rules and its subcategories but not for other categories that have different category classes.

Cyril

Hello,

This is not possible with current CIS. I suggest you contact support to have a feature request created.

Regards,

Cyril

fainemr

Oh I can do it, I just don't like the method. If I have two attributes it is possible: Create one hidden attribute just for storing the results of the classification and set this in the category class and another attribute that contains the actual value of the attribute for the object and is set manually by a user. Then set CIS to categorize based on the second attribute's value and just disregard classification results in the first hidden attribute.