Implementations of the present disclosure include receiving a task including structured data and unstructured data, processing the task to generate a current task object, the current task object comprising a plurality of task attributes, retrieving a previous task object, for each task attribute of the plurality of task attributes, calculating a task attribute similarity value based on a corresponding task attribute of the previous task object, determining whether a rule is applicable to a task similarity calculation, weighting each of the plurality of task attribute similarity values using a plurality of weight values, wherein a first weight value applied to a first task similarity value is different than a second weight value applied to a second task similarity calculation when determining that the rule is applicable to the task similarity calculation, calculating a task similarity value, and providing the task similarity value to be displayed on a display.