Determination of receiver in MS/OS-IHS

When capturing items in-house, the correct recipient must be identified or selected on the basis of a few partially incorrect or incomplete entries. In CodX PostOffice, the input or reading of the address is done by the modules MS-IHS or OS-IHS.

The individual steps for identifying the correct receiver are described here. The operation of the two modules MS-IHS and OS-IHS is described in the corresponding online help.

Loading the IHS cache

The IHS cache is an internal memory that stores all receiver data. The recipient data is divided into so-called keywords. The keywords are single words that occur in names, first names, department names, and so on. The keywords are stored in a huge list together with other information.

The IHS cache is loaded when CodX PostOfffice is started and can take a few minutes depending on the size of the database.

Entering a shipment

When a shipment is entered, the processes below are carried out:

Creation of tokens

If a consignment is read in with MS-IHS or OS-IHS, the input is processed accordingly.

The input or what is read in is first divided into individual words (tokens). All words that are separated with space or any other separator are divided.

Example:

Address:
CodX Software AG
Sinserstrasse 47
6330 Cham

Token: CodX, Software, AG, Sinserstrasse, 47, 6330 Cham

The individual tokens are all treated the same from this point on. At this point, CodX PostOffice has no information which token represents which content.

Filtering the tokens with the blacklist

In a second step, the tokens are compared with the words in the blacklist. The blacklist is a list of words which should not be considered for the search of the receiver.

The comparison of the blacklist is done with a similarity search. For each word in the blacklist a corresponding error factor can be specified. The higher the error factor is, the more likely a token is discarded and not included in the search below. So this error factor must be set accordingly.

Search and weight tokens in the keyword list

In the next step, the complete keyword list is searched with each token. A similarity search also takes place, whereby the error factor per keyword group can also be set.

The result of this similarity search is a quality factor for each keyword between 0 and 100%. This quality factor tells how well the token matches the keyword.

All keywords that have too low a quality factor are discarded. These are no longer considered for the selection of the recipient.

The remaining keywords found are then multiplied by the set weighting. This results in the score of the keyword.

For each entity (person, logistics unit, cost unit), a keyword found is only considered once, and the keyword with the highest score is used. highest score is used.
Example: For a person with the keyword "Peter" in the first name (weighting 30) and in the last name (weighting 40), only the hit on the last name is used. on the last name is used.

Merging the keywords

Since there can be one or more recipients (person, customer, cost center, logistics unit) behind each keyword found, these are now combined.

With the weighted keywords, a quality factor (FullScore) is now calculated for each recipient using a special formula.

Now all recipients are removed from the remaining list, for which the FullScore is too low. The best candidates remain.

This is now the result of the recipients found, which are displayed in the list of MS-IHS or OS-IHS.

Keyword groups

The individual keywords are grouped together. The list below shows the keywords that can be searched for and their group.