| Abstract |
The fundamental information unit in the current Web, the WWW, is a node which currently is a document (or a page) that is connected to other nodes using manually created referential links. By creating infrastructure for processing, storage, communication, organization, search, and presentation of all information available on this Web, the most accessible and ubiquitous source of knowledge and communication in human history has been created. The Web created a cyberspace in which people communicate, do research, shop, and entertain. The impact of cyberspace is clear from the fact that in less than a decade a company rose from its foundation to becoming the most influential and wealthiest company on the planet by following its mission: Google's mission is to organize the world's information and make it universally accessible and useful. Of course the entire world?s information has to be in the cyberspace. The unprecedented impressive progress by the Web, however, is only the proverbial tip of the iceberg. Most of the information and knowledge in the world is captured and stays in the form of experiences in different sensing modalities. Technology tamed knowledge in text because in text the experiential data is converted to symbols by humans. Converting sensory data to symbols in computer systems has been exceedingly difficult so far. Also, humans are extremely adept in dealing with sensory and symbolic information in the same situation. Current computing systems face huge semantic gap - they are either designed mostly to deal with symbolic information or with signals. Most emerging applications, however, must deal with information in the real world as well as in the cyberspace. This paper presents EventWeb to organize and access experiential knowledge as easily as textual knowledge, thereby linking the physical space with cyberspace making them work as one. This starts with the following simple conjecture:Suppose that we create a Web in which1. Each node is an event or object2. Each node may be connected to other nodes using a. Referential Links: similar to common links that refer to other related information.b. Structural Links: showing spatial and temporal relationships with other events.c. Causal Links: establishing causality among relationships.d. Relational Links: giving similarity or any other relationship.3. The links can be created by the creator (human or a machine) of the node, can be automatically created, or can be created by other people.4. The nodes of this Web may be linked to the Nodes of the current DocumentWeb, emerging Semantic Web or any other evolving Web.In this paper we explore implications and technological challenges to build such a Web that we call EventWeb. EventWeb organizes events, relationships among events and objects, and all data associated with events. It uses audio, visual, textual, and other data to provide an environment for naturally experiencing the event from the user?s perspective. It analyzes raw data (documents, sensor feeds, ...) to recognize objects and events using the context of the data and all knowledge available on the Web. EventWeb will also reorganize events to satisfy different viewpoints and naturally incorporate new data types-dynamic, temporal, and live data. EventWeb?s evolution will occur in two major overlapping stages. In the first stage, users will create events and link them. A major driver of this phase of EventWeb will be a new generation of Mobile Personal Devices (MPD) that will evolve from the current mobile phones. Sensor networks and pervasive environments will drive EventWeb?s evolution further in its second phase - the Sentient EventWeb. Sensors (event detectors) are placed where events are likely to occur and need to be detected. They measure some physical characteristics that signal events. Combining such micro-events from individual sensors through a web can detect important events. Emerging sensor networks and pervasive environments that detect micro-events will facilitate the next stage in EventWeb?s evolution. This sentient EventWeb will allow automatic detection of real-world events using sensors and aggregation of these micro-events into meaningful events in diverse applications. Because emerging applications demand more real-time situation modeling and response, EventWeb?s growth will likely be more rapid than the WWW?s. EventWeb will offer multiple perspectives of important events. It will permit remote personalized participation in live events such as meetings, lectures, concerts, and sports. A user could archive these events to experience later, maybe from a different perspective each time. EventWeb builds on three fundamental transformative emerging concepts and research areas associated with them in computing: the use of events and objects (rather than objects alone) to model computing to address the needs of cybereal applications; system software architecture and allied system technologies for realizing EventWeb leveraging the current Web technologies; and unified consideration of data, text, audio, video, and other sensory data to compute with holistic experiences. In this paper we will discuss representation of events, operations and transformations associated with events, and the infrastructure required to build the EventWeb. We will present these using two applications that we are building using this approach. |
