To view the Classification Intelligence reporting dashboards you need to be assigned to either the Application Administrator or Records Manager role in Records365.
The Records365 Classification Intelligence add-on requires an additional subscription. If the ‘Intelligence’ menu item is not visible in the left navigation pane, then the Classification Intelligence add-on is not active for your organization. If you believe your organization is subscribed, or would like to enquire about subscribing, please contact support.
Records365 provides a set of dashboards to display metrics for the Classification Intelligence (CI) feature set. Their purpose is to provide administrators with insights into how the feature is being utilized to facilitate governance of semi and unstructured content.
The CI dashboards can be accessed within the Administrative section of Records365.
- Click on the Settings icon in the top right-hand corner of Records365.
- In the left-hand navigation pane click on Dashboard from within the Intelligence section.
The Overview page of the dashboard displays some key metrics to an overview into how CI is being leveraged by Records365 to apply policy to records.
The first set of numbers indicate how many records have been categorized, and what portion have been categorized by CI. The total number includes rules-based assignment, manual assignment via reschedule or physical record assignment and finally assignment through the use of CI. Records which have a suggested category from the CI feature but are not yet accepted are excluded.
The subsequent group of numbers shows how much of your file plan is covered by CI. It highlights how many of the total file plan categories are included as part of the currently active machine learning model. For more information on including categories into a model see the documentation section on Training.
Classification Acceptance Rates
The Classification Acceptance Rates charts provides an overview of what percentage of suggestions are accepted over time. This is indicative of the overall performance of the currently active model.
A rejected suggestion is specifically when a user performs a Reschedule operation from the Intelligence page. It is worth noting that Rescheduling from any other pages in Records365 is not treated suggestion rejections. For more information on applying records categories using CI see the Managing Suggestions documentation.
The information in this chart makes it really easy for you visualize your model’s performance over time. The higher the percentage of acceptance rates, the better the model. Model performance, and therefore acceptance rates, may be improved with continual training using content that is most representative of each file plan category.
The Classification Breakdown chart shows what portion of content is being categorized using CI over time. It demonstrates what percentage of content has been categorized by CI as opposed to metadata-based rules. All unclassified content, including records that have pending suggestions, is excluded from this chart.
With the right level of investment into training and curating content you should see an upward trend in the ratio of content classified using CI. As this upward trend occurs, metadata-based categorization can be optimized to strike the right balance between metadata and AI-based categorization to give your organization complete coverage.
Record categories can be applied to content using features other than metadata-based rules and CI, such as rescheduling and manual classification for physical items. The above graph only tracks CI and metadata-based categorization, so you may notice that some segments in the graph may not reach 100%.
The history tab is where you can view the results of all training runs that have been carried out in Records365. See the Training page for more information on training a model.
As illustrated below, you can see the following information about the model produced from a training run;
- Which model is active, and is currently being used to offer suggestions to records
- The user who initiated the training run
- When the resultant model was created
- How healthy the resultant model is
- The status of the model, the latest successful model will be promoted to have an Active status.
Clicking on a model will display additional information about that particular item, like the lifecycle and the number of categories included in the model.
Please find links below to other documentation articles mentioned in this page