Browse Prior Art Database

Method and System for interactive multidimensional sorting Disclosure Number: IPCOM000202027D
Publication Date: 2010-Dec-01

Publishing Venue

The Prior Art Database


In data collections with several statistically independent attributes the ability to sort by different attributes is helpful. For example, when locating a particular document for which the exact timestamp is known, sorting by timestamp will make it easy to navigate to that document. Similarly, when knowing the exact size of a document it is easy to locate that document after sorting by size. In contrast, when the attribute values of a document are only vaguely known, combined sorting may help. While the usual way to do combined sorting is to use hierarchical sort order, in this document a different approach is proposed to allow combined sorting by several attributes in interactive scenarios.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 29% of the total text.

Page 01 of 14

Method and System for interactive multidimensional sorting

Consider a data collection with several orthogonal attributes; as an example, let us assume a collection of documents with attributes age and size. Prior art systems allow sort by each of the attributes (sort by date or sort by size), and maybe hierarchical sort by several attributes (e.g. first sort by age, then sort all documents of same age by size; or, first sort by size, and then sort all documents of same size by age). These sort strategies are not sufficient for all scenarios. As an example, consider a scenario where a user is looking for a document in a collection of 1000 documents, with the knowledge that that document is among the 10% of largest size and within the 10% newest age. Sorting by age would still require to look at documents #1 through #100 (on average, this requires to look at 50 documents). Similarly, sorting by size would require to look at documents #1 through #100 (on average to look at 50 documents). Hierarchical sort by two columns would not give much relief, in particular when the document size and the document age would have many distinct values. Proposed is an interactive, combined sorting method that allows to factor in several attributes in a symmetric way. This means, the effective sort order is a ranking order merged from several attribute search orders, thereby allowing to compensate a low ranking value in one attribute by a high ranking value in another attribute. Preferably, the contributions of different attributes can be adjusted dynamically in an interactive way, where the user can interactively adjust the balance between the contributions of several attributes.

Known solutions:

Sort by a single column. Disadvantage: user's knowledge about sort order in other columns cannot be leveraged.

Hierarchical sort by multiple columns (first column dominates, low order columns then do only sort within same values of higher order columns)

Ranking formula like in search engines. In prior art systems, such a formula cannot be adjusted interactively by end users, and the resulting sort order is not precomputed for alternative values.

New solution:

Provide a user interface that allows dynamically to adjust the weight for attributes. Preferably, a user interface control (e.g. scrollbar or slider) is offered for that purpose.

Preferably, that user interface control is calibrated to allow only a small number of distinct values, for example, 10 steps from the lowest to the highest value. In these cases, the resulting order can be precomputed and rendered dynamically and interactively while the slider is moved.

In the following, the behaviour is illustrated using a fictive set of documents, with the following details:
the document set consists of 1000 documents
there are two attributes that can be used for sorting; date and size
the columns date and size are assigned by a random generator
the columns size and date are not correlated
as document of interes...