Flexible Querying of Personal Information

The amount of personal data that users store and access in personal information systems has grown massively. Simple flexible search tools are a necessity for users to be able to access the data they need. In addition, the file organization that is typical in file systems is not a practical logical view anymore as applications may store information across files, or combine different objects into a single file. Users cannot be expected to know or remember the exact location and specific details about the data they are looking for; search techniques should allow for some approximation during query processing in order to return useful results while accounting for possible errors in the query. We present a scoring framework that takes into account approximation in both the content and structural dimensions of the data. Our query model extends the search to consider information across file boundaries. We implemented a prototype of our search techniques over a real data set and report on our experimental results.