Vues :

Question:

What do I have to keep in mind when setting up the Intelligent Indexing Service?


Answer:

For setting up the Intelligent Indexing Service please note that on the one hand it is important to train the service correctly but on the other hand a proper mapping is mandatory as well. In order to avoid errors from the outset you should think about a reasonable use of Intelligent Indexing and keep the following details in mind when setting up the service.

Planning phase and basics:

  • The Intelligent Indexing Service primarily operates with templates. A template is a particular type of documents with strong similarity. It is for instance a set of delivery notes of a parts supplier or the invoices of a telephone company. The similarity of the documents is the key factor, as the whole service is based on it. Therefore, it is not working with altering emails and other formless documents.
  • Mapping is the assignment of Intelligent Indexing fields to DocuWare fields of a store dialog during document tray configuration or since DW 6.11 on store dialog level. (More details under the item “Mapping”)
  • For setting up the Intelligent Indexing Service only one single store dialog of a file cabinet is used.  Index proposals will be exclusively provided for this store dialog. You can recognize it by the light bulb in your web client.

Intelligent Indexing light bulb

  • All fields you expect to be filled by the Intelligent Indexing Service need to contain index entries every single time you store a document. The colored bars at the beginning of an index field show that the suggested index data is provided by Intelligent Indexing. (details on http://help.docuware.com/en/#b57860t59016n65577) Sporadically filled fields can lead to false results or even worse, empty fields. For this reason, it is crucial to analyze which fields are absolutely necessary for every document type in the planning phase. Only these fields should be mapped with Intelligent Indexing fields. The basic principle is: “Sometimes, less is more.”
  • There are a few mandatory Intelligent Indexing fields in document tray configuration (marked by *). If you do not need a mandatory field, however, you still have to assign it and delete the suggested value in the store dialog manually. For every document type you need to repeat this action manually to train the service. In the future, Intelligent Indexing will not provide any indexing proposal for this field and document type anymore.  Indeed, this is not a recommended approach as it can negatively impact the performance of the service.
  • The Intelligent Indexing Service only observes the first 5 and last 5 pages of a document.
  • The Intelligent Indexing quota is reduced every time a document is moved to the Intelligent Indexing document tray. Therefore, you should only import documents to the Intelligent Indexing document tray which are supposed to be indexed by the service.
  • In case you would like to increase your quota, just send a formless email to orders@docuware.com. This adjustment can be taken monthly.

The Mapping itself:

  • Please take care of the right assignment of the field types:
    Intelligent Indexing text field --> DocuWare text field
    Intelligent Indexing date field --> DocuWare date field
    Intelligent Indexing numeric field --> DocuWare numeric field

     
  • For every file cabinet only one store dialog can be used for the configuration of Intelligent Indexing.
  • In general, an identical mapping for all document trays of a store dialog is required.
  • Intelligent Indexing can be applied for storing to several file cabinets. If you want to store a specific document type or document types with great similarity (regarding the vocabulary) to different file cabinets, you need to make sure that the assigned locations in the document are always the same. (see illustration) As already mentioned please take care of an equal mapping of your document tray configurations of a store dialog and make sure that all mapped fields are filled. The field names of the store dialogs can differ but the field types must be identical.

Intelligent Indexing same document types
Completely different document types can be stored to different file cabinets by using unequally mapped store dialogs as well. (see illustration) The focus must be on the great dissimilarity of the document types (regarding the vocabulary).

Intelligent Indexing different document types

  • The subsequent adding of Intelligent Indexing fields to the mapping is not possible and should not be performed.
  • The Mapping may not be modified after storing documents without having reset the Modelspace.
  • It is possible to remove an Intelligent Indexing field from the Mapping if it is not needed any longer.
  • The Intelligent Indexing Service does provide exact date but not any time for datetime fields. In this case it is advisable to use DocuWare date fields.
    If one of these points have not been taken in consideration and Intelligent Indexing shows extremely weird behavior, we recommend a correction of the false configuration and a reset of the Modelspace. Afterwards all document types must be relearned.

Learning document types:

Miscellaneous:

  • A high document resolution is very important for Intelligent Indexing. For optimal results a resolution of at least 300 dpi is advised. The same applies for self-scanned documents. Please take care of optimal scan settings.
  • Field filters can be used to prefilter Intelligent Indexing suggestions for DocuWare fields. Only index entries which are listed on the field filter document will be displayed in the DocuWare field. Field filters can be added or adapted subsequently but should be used with caution.
  • Intelligent Indexing does ignore normal select lists!

For further information on this topic you can visit help.docuware.com. For optimal search results please use Microsoft Internet Explorer and Google Chrome.