1. The Classification Algorithm
Clocktimizer's algorithm reads through your time card narratives and classifies the timecards based on the activities it has been trained to recognize in those narratives. This is the core functionality of our product and what our data is based on. The algorithm has a supervised learning functionality that is driven by your feedback. This feedback is used to improve the recognition quality of the algorithm across Clocktimizer's customer base.
How do I provide my feedback on the classification algorithm?
If you see an improvement that can be applied to the entire algorithm e.g. an activity is missing, an acronym is not classified or a commonly used abbreviation is classified incorrectly, let us know by sending an email to email@example.com. We will review your feedback and get back to you with updates.
2. Activity Grouping
The activity grouping is a feature that allows you to group the Clocktimizer activities. Our customers use this to roll up the Clocktimizer activities into a more birds-eye view. The core algorithm remains the same, but you can create several groups with (UTBMS) codes, phase codes or client codes.
These activity groups can be applied throughout Clocktimizer (e.g. on a matter level, as a budget, as a comparison feature or for generic analysis). You can have multiple activity groups, for example, one per practice group, or specific activity groups per client. The activity groups remain within the Clocktimizer environment of your firm.
How do I group Clocktimizer activities?
Please see the Quick Guide on activity grouping in our support portal.
Tip: we have templates available for the UTBMS M&A Codes, UTBMS Litigation Codes, and Jackson Reform Codes. Contact firstname.lastname@example.org and ask for the templates to be uploaded in your Clocktimizer environment.
3. Manual and Mass Categorization
The manual categorization enables users to change the classified activity on the time entry-level to assign those last remaining unclassified entries to the correct activity. The general algorithm is not updated, rather the specific time entry is updated. The updates are applicable to your firm's Clocktimizer only. In essence, you can create your own taxonomy in Clocktimizer. The manual and mass categorization remains within your firm.
How do I manually categorize time entries?
Please see the Quick Guide on Manual Categorization in our support portal.
How do I manually categorize multiple time entries?
Please see the Quick Guide on Mass Categorization in our support portal.
4. Machine Learning
To be complete (but if you are overwhelmed already, please forget this), Clocktimizer can apply its machine learning engine to your time cards. This is an engine that is not related to the above. The machine learning engine allows you to determine a predefined number of buckets in which the engine should classify a set of time cards.
Once you provide Clocktimizer's machine learning engine with a dataset that contains roughly 5,000-10,000 time cards with the correct bucket assigned to each time card (the training data), the engine will build a predictive model that will assign future time cards to these predefined buckets. The key here is to have that training data available. We are happy to tell you more about this!