 |
 |
 |
|||
|
EVENTWEB FOR DISSEMINATING CULTURAL HISTORY Ramesh Jain 1. Introduction The World Wide Web (WWW) has created a platform for use by people to create as well as consume information. By making information easily accessible to everybody who cares, this has brought a major information and knowledge revolution in our society in the last two decades. Documents existed on computers that were connected using Internet much before WWW came into existence. All documents on the Internet were independent. Berners-Lee thought that documents could be interconnected by creating explicit links between them. By creating tools and an environment that let developers and end users easily link, create, and access documents, Berners-Lee created the Web. Once browsers like Mosaic were developed, the Web became one of the most powerful information and knowledge device human civilization has ever developed. Today’s Web, however, is a web of documents - that is, it assumes that all information and knowledge exist in documents. The basic unit of these documents is a page. Photos, videos, audio, and three-dimensional objects must be part of a page to exist on the Web. In creating the Web, developers focused on developing a language, HTML, for building and presenting pages. There are now tools to present other types of data, but pages remain central to the current Web. The concept of linking documents explicitly isn’t new - writers have long been doing so through footnotes, bibliographies, and references to other documents. Hyperlinks let you link an electronically stored article to another. The Web, however, takes this to a new level. Web page authors create links between documents on the Web to refer to a document explicitly that they consider relevant in that context. These documents could be located physically on any computer, but the infrastructure allows them to access effortlessly just by clicking on the link. I call these links referential links, or r-links. R-links are usually one-directional—that is; the referenced node doesn’t necessarily point to, and might not even be aware of, the pointing node. Ultimately, what makes the Web “the Web” is links between documents. The entire Web can be viewed as one document due to the innovative technological infrastructure that has been built in the last two decades. 2. Current Status Almost 20 years have passed since Berners-Lee worked on the idea of linking all the information and knowledge available in cyberspace. Rapid advances have resulted in some striking tools and technology: • Crawling: Web crawlers find all the information created by anyone and posted on the Web. No centralized registration is required. Crawlers will discover a document if you put it on a publicly available Web server. • Search: Search engines find the locations of all pages in which a set or combination of keywords appears. Developers are continuously refining the technology to analyze users’ intent and bring in related pages. Search engines also have made progress in ranking relevant pages using sophisticated approaches. • Multimedia: The amount of visual (images and video) and aural (speech and music) information on the Web has been rapidly increasing. The proliferation of digital cameras and mobile phones has been a major source of such multimedia information, which is likely to become as common as text, and possibly dominant, in the next few years. In the longer term, multimedia will dominate. • Sensors: Many types of sensors—from simple temperature sensors to sophisticated webcams—are becoming part of the Web. The trend is likely to continue so that we’ll see significantly more sensors than pages on the Web. These sensors will be creating continuously new data, and many of these applications will maintain a record of this data. • Mobile phones: Mobile phones are a big driver of the next generation of computing and Internet technology. Just as PCs revolutionized computing by extending a computer’s reach to many different populations, mobile phones are taking computing to emerging countries where millions are illiterate or uneducated. Moreover, because mobile phones are always with their users, they’re significantly more influential than PCs. Finally, today’s mobile phones come with built-in multimedia and sensory features. 3. Weaving a Web of Events Suppose we create a Web in which each node is an event. Each event has informational and experiential attributes and links describing its structure, causality, and relationships. Informational attributes describe the event and its participants using text or other attributes such as a Social Security number or telephone number. Experiential attributes associate sensory data such as image, video, and audio for the event. These sensory data, synchronized if necessary, present the experience of the event. Although aural and visual characteristics are common, you can associate other sensory information and render it using aural or visual mechanisms. For example, you can capture some data using sensors that aren’t meaningful to humans and create medical or weather images to show humans the data in visual form. Many types of links could exist among events. Initially, the events’ authors will explicitly create most links. Many of these links will be referential, as in DocumentWeb. Event creators will think of past and upcoming related events and might explicitly link those events to theirs. In the EventWeb, event authors might create links to documents and experiential data after the event has passed. Often these links will be related to reports on the event. Also, people who participated in the event will likely create explicit referential links. So, links in events might differ from document links - most event links might be owned by people other than the event creator. In addition to r-links, you could use c-links to represent causality, S1-links to represent sub-events, S2-links to represent super-events, and t-links to represent similar events. You could extend event types depending on the application. EventWeb attempts to capture multiple types of relationships through different types of links. You could easily represent the links’ semantics in the system as the link type and use them for different types of traversals. All these links don’t need to be created automatically. In the WWW, r-links are created manually. In EventWeb, you could create some links manually and others automatically. A fundamental difference between a DocumentWeb node and an EventWeb node is that in the latter, each node is solidly grounded in the real world using the event’s physical location and time of occurrence. In cyberspace, where the DocumentWeb lives, time and location are secondary. We can see two additional fundamental differences between the DocumentWeb and EventWeb: • The documents composing DocumentWeb are created by humans and thus represent subjective descriptions of objects and events. EventWeb, however, relies on experiential data, which you can capture by placing sensors in an environment where events are taking place. • Events are naturally linked to other events. An event might cause one or more events and might also result from a sequence of events. In fact, much of the research and analysis in most fields is essentially the discovery of links among events. Disciplines ranging from history to human physiology study relationships among different events occurring at multiple granularities and at different locations. Despite these differences, a strong DocumentWeb will persist and will be tightly linked to the rapidly evolving EventWeb. DocumentWeb represents much existing human knowledge in textual form. As a persistent version of speech—the most powerful communication mechanism among humans—text is a powerful representation mechanism for human knowledge. For a long time, documents were the only way to capture and communicate knowledge. Technology’s rapid evolution in the last few decades is changing this. Now audio, photos, and videos are becoming dominant medium for capturing, storing, and distributing knowledge and information. However, text will remain a primary mechanism for knowledge representation and communication. DocumentWeb and EventWeb are different from the Semantic Web, Web 2.0, and other such terms Web communities commonly use. Both researchers and user communities use these as special extensions of the current DocumentWeb and will extend these equally well to EventWeb. DocumentWeb and EventWeb describe a node’s fundamental character, whereas the Semantic Web and Web 2.0 describe tools and techniques for facilitating Web functionality. 4. Objects and Events Philosophers have been exploring the relationships among objects and events for a long time, but they have yet to resolve any clear differences or relationships[1]. Both objects and events play fundamental roles in the dynamic world we live in. Objects are better at capturing persistence, whereas events are better at capturing dynamics. Objects participate in events, and events describe transformations in object attributes. Events also serve as a powerful semantic abstraction of time. Calendars help you structure time, but they don’t represent an abstraction of time; events do.
Figure 1: Six facets of an event, as discussed in the event model Representing even the simplest event requires using a tractable form of spatiotemporal logic as the system’s semantic underpinning. Traditional formalisms, such as the situation calculus, inherit the inherent intractability of first-order logic. Although researchers have made some progress in developing description logics [Lut07] with time as the concrete domain, similar efforts in the spatiotemporal concrete domains are still in their infancy. If events are to serve as the fundamental representational and organizational mechanism for multimedia experiential systems, we need strong conceptual, engineering, computational, and human-centered designs. As Figure 1 shows, an event defines the following aspects [Wes07]: - Information about the event, location, and time; We can combine events to define other (compound) events. By providing flexible and expressive mechanisms to define these components and their associated methods, we could define events at different granularity and complexity levels. The event environment should provide tools for defining events of interest from many disparate application domains. You could build the universal concepts of time and space (starting with broadly adopted time and space ontologies) and associated reasoning (e.g., nearby, before) into the computational fabric to enable more powerful event handling [Per07].
Figure 1: Six facets of an event captured in an event model. 5.0. From Events to EventWeb Companies like Eventful, Upcoming (part of Yahoo!), and Zvents are using Web 2.0 technology to build an environment for events. New tools will let users submit events, import events to and from calendars, request specific events, find out who will be attending an event, and so on. Such tools will contribute toward building a powerful calendar of events at the local and global levels. There’s an interesting irony, however. Although most people are using Web 2.0 tools to create events, these tools aren’t creating a Web of events. In current systems, each event is a node and isn’t connected to any other event. These unconnected events are like the files on unconnected computer systems before Berners-Lee envisioned the WWW. Connecting files resulted in applications that were unimaginable before the Web. We have much stronger reasons to connect events than documents. Natural relationships exist among events at the structural, causal, and similarity levels. But Web 2.0 tools won’t give us a Web of events. And by now the power of networks—that is, the power of connections—is obvious. EventWeb will let us create these connections. 5.1. Components of the EventWeb The EventWeb architecture is a natural evolutionary step from the current WWW and hence will use several components. Events bring several new challenges. The Figure 3 shows EventWeb’s key components, which we briefly discuss here.
5.2 Event Entry Systems An Event Markup Language (EML) and a good authoring/production environment will let people post their events and related information and experiential data in the form of photos, audio, videos, and textual data. EML will provide an environment for expressing and creating relationships among events. 5.3 Event Capture Systems When sensors are used to capture events, most events will be automatically captured using sensory (say, video) data processing. Many applications might initially use semi-automatic event entry, which will assist automated analysis in capturing and indexing subevents. 5.4 Event Archiving Events have an interesting life cycle. They’re planned, they take place, and then people store their experiences of the event in the form of experiential data and relate them to past and future events. Interestingly, past events play a more important role in our lives than current and future events. For example, all sciences rely on the analysis of past events. Most paintings, novels, movies, and news reports relate to past experiences. An environment for capturing and archiving events, along with associated metadata, is therefore essential, as is the facility to continuously add new information and experiences and links to these events. 5.5. EventWiki and Social Web A key transformation taking place in computing is the rise of a participatory culture in the digital world, where it is socially "cool" to contribute content. Wikipedia, the blogosphere, YouTube, MySpace, and Facebook are now integral parts of today's digital culture. A quick analysis of popular participatory sites shows that much of the participation centers on personal and community events. We will further this trend by creating an EML-based EventWiki. 5.6 Event Crawling When billions of events (past, current, and future) exist, how do we find them and correlate them by their temporal, spatial, and thematic aspects? An event environment should let users discover events of interests, and then communicate with others to plan future events or analyze past events. 5.7 Indexing and middlewareIndexing of multidimensional data sets has been extensively studied in the database and multimedia literature. Typically, such systems model multimedia content using a set of features and each of the features maps to a vector in a multidimensional space. In an eventweb, one must develop novel indexing techniques to consider space, time, informational, and experiential data. 5.8 Event SearchEvents are inherently multidimensional. Mechanisms for browsing and searching for events will likely differ significantly from those used in the current keyword-dominated WWW environment. How do we query about events and explore them with integrated spatiotemporal-thematic processing, with built-in support for proximity, similarity, causal relationships, and contextual knowledge (e.g., application of geographical, political, and other context)? Event search and browsing become a challenging problem because we must develop techniques to parse event queries using both the system and personal ontologies. 5.9 Presentation and Interaction Presentation and interaction environments in DocumentWeb are keyword boxes and ranked lists of results. These aren’t adequate even for DocumentWeb but are acceptable because of text’s linear nature. An effective interface in EventWeb will be a geospatial and temporal interface, far ahead of the current state-of-the-art. Moreover, the EventWeb interface should have both the information richness of a full-blown geospatial system and the interactivity and flexibility of a computer game to immerse and engage the user in navigating and exploring the experiential environment of EventWeb. 6.0 Seraja: The first Version of EventWeb We are implementing early ideas related to EventWeb in Seraja (seraja.ics.uci.edu). Seraja builds on the event model described in[2] and provides tools for creating an event; adding informational, experiential, spatial, and temporal data related to event; and creating structural and causal relationships. It also allows search based on time, location, tags, keywords, and any material related to photos and videos, publishers, and type of events. These functionalities are in their early stages but are usable. We are using Seraja to build an experiential multimedia management environment to create regular events at UCI and other places, and for many other events by our group members. We are also applying it to cultural history and other applications. The system will be put in regular use by a community of users by May 1, 2008. The system is based on donated software, for which we have all rights for research and making open source. We will refine, enhance, and enrich the software based on user feedback and research results. We will then make it available as open source software to get feedback from users in disparate application areas for understanding requirements in building EventWeb. Seraja can be a tool to help Buddhist groups collect and deal with information that extends over time. For example, the planning of a major public event, the study of the history of a monastery, or the description of the life of an important person. These tools can be easily used by people who can use a computer now and will be made available for Mobile phone users in the near future. We hope to assist groups in the future as they move toward new ways of communication. [1] A. Quinton, “Objects and Events,” Mind, vol. 88, no. 350, 1979, pp. 197–214. |
 |
||||
 |
|||||
