Cluster similar documents with HighQ AI
Contract clustering uses AI to group similar documents. This feature allows you to identify contracts that are, for example, part of a series of revisions or branches of the same original contract, and helps to clarify the history of the signed and executed contract.
Similar to contract deviation analysis, HighQ AI bases clustering on a deviation score. You can adjust the acceptable deviation to reflect changes that are typical in your workflow. You may organise clustered files as required, for example, storing signed contracts in a contract 'database' for quick reference and security.
Clustering similar documents can also find duplicate copies of a contract and allow you to easily assign a contract template to further analyse revisions.
note
Please do not use clustering on sites that contain placeholder files, as they can interfere with the clustering process.
Opening the Clusters window
If you are a site admin, select
Admin
, then under AI Hub
choose Clusters
to open the Clusters
window. If this option is not available, check your AI Hub configuration to enable file clustering.
note
When you first open the Clusters Window, the list of files will be empty
Setting a deviation level
Adjust the deviation level to control how similar the files must be to be grouped. If you select a deviation that is 0, or close to 0, then the files must be very similar, if you choose a higher number then more variation is allowed.
note
For most purposes, start with a deviation level of 30; this should indicate how varied your files are. You may adjust the value after the first analysis.
Select
Save and cluster
to start the analysis.
A
Cluster
window confirms how many files will be analysed. Select Cluster
to continue.
note
All Word and PDF documents on your site will be analysed. You can choose to ignore a selection of documents after they are added to the list of clusters.
As the analysis may take some time, a status message is displayed in the top navigation bar:
If you need to stop or change the analysis, select
Cancel
in the Clusters window. You may then change the deviation and restart the process.
When the results are ready, a list of documents is shown.
Click on the arrow next to a clustered document to see the files in that cluster, each with a reason and a percentage score for the match:
note
The top-level 'representative' file is often the signed contract
The
Template
column shows which standard template the file is linked to.note
Contracts and templates must be in Word, PDF or TXT format. If a Word document is compared to a PDF equivalent, then the match may be lower than expected (i.e. more changes) as differences in the underlying file formats may be included as changes to the text
Exact matches
If a cluster includes an
Exact match
, then it is likely that the files are identical. This allows you to manage duplicate files.
New documents
Select
Save and Cluster
again to include any documents added to the Files module since the last analysis.note
This also removes any manual changes; except documents that have been ignored (see below), which will not be included in any future clusters
Viewing a document from the Cluster window
Click on a document to open it in the
document viewer
. Additional actions are available in the document viewer window:
Actions in the Cluster window
Select one or more files in the
Cluster
window and select Action
to manage the files.
- Move or copy- move or copy the file to an existing folder in the Files module
- Set as representative- select a document then choose this option to define that file as the representative document
- Start new cluster- move the document out of a cluster. It moves to the top level in the cluster list
- Move to another cluster- select to move files that have not been clustered, or have been clustered with inappropriate files, to a different cluster. Select a cluster and selectMove to other cluster
- Ignore this file- remove the selected document from the list of clusters
- Assign contract template- assign a template to the document. SelectSave and analyseto apply the change
note
If you select the representative document for a cluster and then
Assign contract template
, all documents in that cluster are also assigned that template.note
If a document is moved, or the representative document changes, then the percentage match will be reset to zero as no comparison has been made for the new relationship:
System and site admin
If the AI Hub is enabled on your instance, Clustering is normally enabled by default for the instance, but is off for each site. If you want to enable or disable the feature, or check the settings, both system and site options are provided.
note
If clustering is not available on your instance, check with HighQ support.
System admin
The HighQ AI Hub must be enabled and
Enable file clustering
must be ON
in the Third party services
section of System Admin
, System Settings
to enable file clustering on your instance.
Site admin
At the site level, enable or disable
AI file clustering
in Admin
, AI Hub
, Configure
, Advanced settings
.