No abstract

This paper presents an iterative method for generative semantic clustering of related information elements in spatial hypertext documents. The goal is to automatically organize them in ways that are meaningful to the user. We consider a process in which elements are gradually added to a spatial hypertext. The method for generating meaningful layout is based on a quantitative model that measures and represents the mutual relatedness between each new element and those already in the document. The measurement is based on attributes such as metadata, term vectors, user interest expressions, and document locations. We call this model relatedness potential, because it represents how much the new element is related and thus attracted to existing elements as a vector field across the space. Using this field as a gradient potential, the new element will be placed near the most attracted elements, forming clusters of related elements. The relative magnitude of contribution of attributes to relatedness potential can be controlled through an interactive interface.Unlike prior clustering methods such as k-means and selforganizing-maps, relatedness potential works well in iterative systems, in which the collection of elements is not defined a priori. Further, users can invoke relatedness potential to re-cluster elements, as they engage in on-the-fly provisional acts of direct manipulation reorganization and latching of a few most significant elements. A preliminary study indicates that users find this method generates spatial hypertext documents that are easier to read.

Metadocuments are documents that consist primarily of references to other documents, and elements within them. Our active browsing web visualization tool generates an evolving series of navigable metadocument snapshots over time. The granularity of browsing is shifted, from documents to the finer grained information elements, which are metadocument constituents. The program conducts expression-directed automatic retrieval of information from the web. It performs procedural visual composition of the information elements to form spatial hypertext. The user can express interest and design intentions through direct manipulation interactions with the visualized information elements. As prior versions of the tool lacked the capabilities of save and load, they were entirely process-oriented. The metadocuments existed only as transient states. This paper is an early report on our new metadocument authoring and publishing capability, and its potential uses. Saved metadocuments can be published on the web. Once published, they can serve both as static navigable metadocuments, and as the jumping off point from which the information space represented by the collected elements can continue to evolve. COLLECTING INFORMATION ELEMENTS INTO METADOCUMENTSUsers of hypertext often need to collect references to significant places that they encounter while browsing. These collections are metadocuments. They are documents that consist primarily of references to other documents, and elements within them [2]. They consist both of references by name, such as image elements and hyperlinks, and by value, in the form of textual quotations. We call the image references and quotations information elements. For each information element, in addition to any embedded hyperlinks, there is always an implicit reference back to the original document, which we call its container. Schraefel articulates the importance to metadocument authors of the connection between an information element and its container [7]; when we collect information elements from the web, we want to be able to easily return to the sources of the quotations. Each information element can be thought of as an enhanced bookmark. EXPRESSION-DIRECTED GENERATIVE BROWSINGOur program [3] generates collections of information elements automatically, over time. The process begins with the specification of seed URLs. These may refer to static or dynamic web pages, including search engine queries. The seed content is decomposed into information elements.The program builds a model of the structure of the information, and of the user's interests [5]. The graph structure of the web is represented, along with a term-based textual index of information elements and document references. Each node includes attributes that record the user's expressions of interest.The program automatically, periodically engages in several dynamic operations, based on the model. It chooses information elements to display, document references to crawl to and download, and regions of the screen for the placement of successi...

combinFormation is a tool that enables browsing and collecting information elements in a generative space. By generative, we mean that the tool is an agent that automatically retrieves information elements and visually composes them. A combinFormation session presents a dynamic, evolving recombination of information elements from different sources. The elements are manipulable in the information space. Recombination is the process of taking previously unconnected elements, and combining them to create new configurations.

No abstract