An accumulated information data store may include topic nodes, each having a text description of limited length and (in some cases) one or more attributes. A particular topic node may be associated as a parent topic node other child topic nodes such that the topic nodes form at least one data tree. An information processing engine may access information in the accumulated information data store and determine that a plurality of topic node text descriptions are similar and classify them as similar topic nodes. At least a part of the text description associated with one of the similar topic nodes may be selected as a favorable text description for the similar topic nodes. The system may also unify the similar topic nodes as identical topic nodes when they are currently grouped together as having the same upper tree hierarchy.