Andrew Wirtanen 9/28/2004 4002-770-70 Intro To XML

The Semantic Web is all about sharing information. A wiki is a website that allows any web user to share or edit data. Therefore, it only makes sense to get a definition from the Wikipedia (http://www.wikipedia.com). Wikipedia defines the Semantic Web as "a project that intends to create a universal medium for information exchange by giving meaning, in a manner understandable by machines, to the content of documents on the Web. Currently under the direction of its creator, Tim Berners-Lee of the World Wide Web Consortium (W3C), the Semantic Web extends the ability of the World Wide Web (WWW) through the use of standards, markup languages and related processing tools" (Semantic Web, 2004, para. 1). In other words, ordinary static HTML web pages display data, but the data has no meaning to the computer displaying the page. With the Semantic Web, meaning will be attached to the data, and computers will be able to use it. With the current state of the Internet, the Semantic Web is not yet technically possible for various reasons such as websites not being compatible and browsers not support all of the necessary technologies. The Semantic Web also lacks a concrete ontology, which is an "attempt to formulate an exhaustive and rigorous conceptual schema within a given domain, a typically hierarchical data structure containing all the relevant entities and their relationships and rules within that domain" (Ontology (computer science), 2004, para. 1). In simpler terms, an ontology is a well-defined group of terms with relationships. Machines sharing an ontology will be able to exchange the meaningful information that they hold. Once both of the technical and conceptual issues are resolved and the Semantic Web is implemented, the Internet should have less of a grey area--all web pages related to a topic should be easily located if one knows what they are looking for. At that time, search engines as we know them will become more powerful because the computers they are hosted on will understand what we are searching for. How is the Semantic Web different from the WWW? According to Tim Berners-Lee, "most of the Web's content today is designed for humans to read, not for computers to manipulate meaningfully" (2002, p. 25). By designing for not only humans, but for computers as well, computers will be able to work together to be able to provide humans with the most applicable and in-depth results. The World Wide Web allows mostly for only one-way communication from the user to a computer (or website). The Semantic Web would allow for two-way communication from users to computers, computers to users, and computers to other computers. The latter would be not a different Web, but an extension of the current WWW with added meaning. How will computers be able to communication with each other? A Universal Resource Identifier (URI) "defines or specifies an entity" (Berners-Lee, 2002, p. 27). Without one defining every entity on the WWW, computers do not know which entities are the same (i.e. soda and pop) and which are different (i.e. soda and beer). That's where ontologies come into play. The idea of a common ontology is nothing new. "Early in the Web's development, detractors pointed out that it would never be a well-organized library; without a central database and tree structure, one would never be sure of finding everything" (Berners-Lee, 2002, p. 26). And that's why search engines do not retrieve all information relative to one's search. Modern search engines, such as Google (http://www.google.com), use spiders (AKA robots or bots) that crawl from web page to web page via hyperlinks (Chamberlain, 2000, para. 3). Sites are indexed (and sometimes cached) by the search engine, and when one searches the sites that are linked to the most that contain one's query are displayed first. If a web page is only linked by a few sites or none at all, one will likely not find the information contained on the page, which could have been very valuable (or useless). This technology is called Information Retrieval (IR), and is "based purely on the occurrence of words in documents" (Guha, p. 702). Two types of searches utilize IR technology: navigational searches and research searches. Navigational searches are directed queries intended to find a particular document. For instance, if one did not remember the address to the How to Fold a Shirt website (http://www.howtofoldashirt.net), one could simply type the phrase "how to fold a shirt" into Google (http://www.google.com) and the site would one of the first results. Developers of the Semantic Web are not interested in these types of searches. Instead, they are interested in Research Searches, where one is using the search engine to find a series of documents that contain information one is looking for. An example of this is if one had to write a 10 page paper on the Semantic Web, and used Google (http://www.google.com) to find websites with information about it. What exactly is wrong with the way Google (http://www.google.com) searches? In his 2003 book Spinning the Semantic Web, Dieter Fensel describes the current state of search engines: Already, finding the right piece of information on the Web is often a nightmare. In searching the Web for specific information, one gets lost in huge amounts of irrelevant material and may often miss the relevant matter. Searches are imprecise, often returning pointers to many thousands of pages (and this situation worsens as the Web grows). In addition, a user must read through the retrieved documents to extract the desired information-so even once a truly relevant Web page is found, the search may be difficult or the information obscured. Thus, the same piece of knowledge must often be presented in different contexts on the same Web page and adapted to different users' needs and queries. However, the Web lacks automated translation tools to allow this information to be transformed automatically among different representation formats and contexts. A related problem is that the maintenance of Web sources has become very difficult. Keeping redundant information consistent and keeping information correct is hardly supported by current Web tools, and thus the burden on a Webmaster to maintain consistency is often over-whelming. This leads to a plethora of sites with inconsistent and/or contra dictory information. (Chapter 1, para. 2-3) IR technology does not work well with the main characteristics of the WWW. The WWW is distributed, so why not utilize this and have different computers host different services? The web is ever-changing, and if search engines are not updated, then links will be out-dated. The web is also massive, and it is hard to index it entirely with bots. Finally, the web is an open world, and there are an endless amount of web pages on it. The largest search engines have only indexed 25% of all of the web pages that exist. The Semantic Web will make it easier to find information from all web pages and sources on the Internet (Fensel, 2003, Chapter 2, para. 3-7). There are five major components to the Semantic Web: XML, XML Schema, RDF, RDF Schema, and OWL (Semantic Web, 2004, para. 5). The eXtensible Markup Language (XML) allows developers to define an arbitrary structure to a document, unlike the HyperText Markup Language (HTML), where the structure is already defined and the developer simply needs to learn how to work with it (Berners-Lee, 2002, p. 26). The XML Schema's job is to simply make sure the XML's structure is correct. The Resource Description Framework (RDF) expresses the meaning behind one's XML by defining the structure and relationships between one's data. Similar to the XML Schema, the RDF Schema gives more meaning to RDF. The RDF Schema describes RDF properties and relationships between them and other resources (Semantic Web, para. 5). The Web Ontology Language (OWL) is the ontology that is used by the W3C (http://www.w3.org) uses in its specifications for the Semantic Web. OWL is comprised of three versions: OWL Lite, OWL DL, and OWL Full (OWL Web Ontology Language Overview). OWL Lite "supports those users primarily needing a classification hierarchy and simple constraints". OWL DL "supports those users who want the maximum expressiveness while retaining computational completeness". And, the least popular and compatible, OWL Full "is meant for users who want maximum expressiveness and the syntactic freedom of RDF with no computational guarantee". The Semantic Web does not yet have an application to browse its XML documents, but modern web browsers can be used to view XML documents in the meantime. Unfortunately, popular web browsers vary in the way that they display the documents and do not conform to W3C standards. Seventy-six percent of Internet users browse the web with Microsoft's Internet Explorer (IE) (Browser Statistics). IE versions 5.0, 5.5, and 6.0 use a Microsoft-specific stylesheet called the eXtensible Stylesheet Language (XSL) instead of the standard eXtensible Stylesheet Language Transformations (XSLT) (Flynn, 2004). Mozilla 0.9 and Netscape 6 both incorporate XSLT and a XML parser called expat (http://www.jclark.com/xml/expat.html) that is more standards compliant. Opera supports XML and CSS on both Windows and Linux, but has bugs similar to Netscape, due to Netscape's influence on the browser. DocZilla is a browser created by a partnership between Finnish company named CITEC and Mozilla. It supports HTML, XML, and SGML, with XSLT and CSS stylesheets, and version 1.0 was released on May 31, 2003. The Defense Advanced Research Projects Agency (DARPA) of the United States Military developed their own markup language called the DARPA Agent Markup Language (DAML) in August of 2000 (Ouellet, 2002, para. 5). DAML, which is used in conjunction with the Ontology Inference Layer (OIL) as part of the OWL, facilitates the ability to use human speech to obtain data from computers. An application of this could be used in search engines; instead of typing keywords, one could type a sentence into a search engine and it would understand what one was looking for. OIL "combines the widely used modeling primitives from frame-based languages with the formal semantics and reasoning services provided by description logics" (Welcome to OIL, para. 1). Known collectively as DAML+OIL, the two languages provide a powerful backbone for OWL for "expressing far more sophisticated classifications and properties of resources than RDFS" (Ouellet, 2002, para. 5). A challenge for the OWL and its subsidiaries is to provide an equivalence relation (Berners-Lee, 2002, p. 28). For example, one may define addresses as containing a zip code, whereas another definition of addresses defines it as containing a postal code. Both zip code and postal code is the same thing, but two computers may disagree unless an equivalence relation is provided by their respective ontologies. Another challenge for the Semantic Web is the ability to be able to distinguish between websites that have accurate information and ones that do not (Guha, 2003, p 702). For example, one could have created a joke website stating that one was president of Malaysia. However, in reality, there is no president of Malaysia because they have a prime minister, and someone else is prime minister. But, a computer program would not be able to sense this without some pre-defined intelligence. Semantic Web researchers have developed semantic search utilities to try to tackle these issues and demonstrate the differences between IR technology and semantic searches. Semantic searches could also work together with IR technology. For instance, a search engine could still provide a list of documents based on its spidering results, but on the side there could be a column with a collection of information gathered from the Semantic Web. The two semantic search tools in development are called Activity Based Search (ABS) and the W3C Semantic Search. Both are built upon of the TAP infrastructure (http://tap.stanford.edu), which provides a query interface called GetData to retrieve information (Guha, 2003, p. 702). GetData returns a graph or graphs that contain the query and all other related information. Each graph represents a different Universal Resource Locator (URL), or web address. Since there is very little semantic data on the WWW, TAP uses its own Knowledge Base (KB). An external registry keeps track of which graphs every URL corresponds to. It also caches, or stores, the responses to GetData requests in order to combat the problem of high latency (Guha, 2003, p. 704). ABS and the W3C Semantic Search are built on different types of data. To obtain data for ABS, HTML scrapers are used that dynamically convert regular HTML pages into machine readable format (Guha, 2003, p. 704). By doing this, the developers can pretend that the external sources are all machine readable. ABS also makes extensive use of the TAP KB. The TAP KB "contributes about 65,000 people, organizations and places, which together cover about 17% of the searches that take place on an average day on the ODP website". ODP stands for the Open Directory Project (http://dmoz.org), and it is the largest web-based human directory on the WWW (About the Open Directory Project). The W3C Semantic Search on the other hand, is relatively much smaller (Guha, 2003, p. 704). It keeps track of people related to the W3C, W3C activities, working groups and other committees, documents and news. Both systems use an ontology from the TAP KB that describe their contents. If semantic searches are going to be included with regular IR searches, then how does the semantic search determine the denotation of the search? Semantic search researchers have developed a methodology to attack this. The first step is to choose a denotation of the search query (Guha, 2003, p. 705). This could be done based upon the denotation that is searched for more often. Alternatively, a user profile could be created in order to be used to determine the best denotation. For example, if the user profile states that the user likes billiards, and the user searches for "pool", then only results relating to the game of billiards could be displayed since they are more likely for this particular user. This could be problematic, because the user could be searching for the nearest community pool that they can go swimming in. The user should still be able to specify a denotation if the semantic search interprets their request wrong. A similar tool to that is a technique used by Amazon (http://www.amazon.com). Amazon makes recommendations for users based on recently viewed items. Similarly, a search could base its results around the user's recent searches. For instance, if the user has been searching for baseball teams and they type "devil rays", then they are most likely to find data related to the Major League Baseball team called the Tampa Bay Devil Rays. Even if the semantic search returned results related to the correct topic the user was searching for, there are still two more obstacles in the way: how does the semantic search determine what to show, and how does it display it? A way to do this is to first do a breadth first walk around the anchor node (the query) and collect a predefined number of triples (a graph with three sets of data, which refers to a URL) (Guha, 2003, p. 707). This search would also have to accept only a certain number of pages from the same source. The problem with this methodology is that the semantic search assumes that URLs with similar triples are similar. This could make the search more vulnerable to spam and irrelevant information. Formatting the data depends on the search interface. Currently, however, the search results would need to be transformed to HTML in order to be accurately displayed on all browsers. To truly understand the potential power of the Semantic Web, one needs to understand the current state of the WWW, in particular its web services. A web service is defined as "a collection of protocols and standards used for exchanging data between applications" (Web Service, 2004, para. 1). Currently, to use web-based services, one needs to specify explicitly which one to use because web-based programs do not have any attached semantics. But, what if they did? What if programs could actively seek out web services based on meaning? Tim Berners-Lee describes this process as "service discovery", and states that this can only happen if both the service and the agent share a common language (or ontology) (2002, p. 29). If a computer could utilize service discovery, then simple tasks may be carried out with ease (Berners-Lee, 2002). For example, one could tell a program on their computer that they are hungry and in the mood for Chinese food. The program, with access to the Internet, would then communicate with one's refrigerator to determine if there is any leftover Chinese food. If so, the program would then communicate with one's microwave to determine the optimal time and power used to heat it up. If not, the program would search online for Chinese restaurants in the area that deliver. The program would also seek out restaurant review sources to rank the restaurants by value. Seconds after the request was made, an electronic menu is displayed on the screen, and one could choose to order from this restaurant, or choose another form the generated list. After an electronic menu is filled out, the request is submitted by the program, and a Chinese food delivery man arrives at one's door 20 minutes later. Not everyone, however, believes that the Semantic Web will be successful. An argument against it is that it is based upon syllogisms, which are "a form of logic, first described by Aristotle, where '...certain things being stated, something other than what is stated follows of necessity from their being so'" and not always true (Shirky, 2003, para. 2). For example, here is a syllogism that is not true: Rochester is a city in Minnesota. Xerox is based in Rochester. Therefore, Xerox is based in Rochester, Minnesota. Even though the first two statements were correct, the third is wrong. How will computers be able to tell which syllogisms are correct and which ones are false? Because of this question, many people, including Clay Shirky believe that the Semantic Web sounds nice, but it will take an unreasonable amount of work to achieve. Shirky claims that: "like many visions that project future benefits but ignore present costs, [The Semantic Web] requires too much coordination and too much energy to effect in the real world, where deductive logic is less effective and shared worldview is harder to create than we often want to admit (Shirky, 2003). In conclusion, the Semantic Web has potential to make our lives easier than ever before by providing meaning to data that computers can understand. With a common ontology, such was the OWL, computers can communicate with one another to share information and services. One of the first ways that the Semantic Web will benefit us is with search engines. Traditional IR technology is limited and problematic, but a semantic search could yield a variety of information by gaining it from graphs known as triples, instead of from keywords. An IR search engine with a semantic component will most likely be developed first, followed by pure semantic search utilities. The ABS and W3C Semantic Searches are already simulating this, but have to rely mostly on their own data to test out their searching capabilities. Meanwhile, skeptics such as Clay Shirky believe that the Semantic Web will never be possible because it relies upon syllogisms, which aren't always true. But, if the Semantic Web can detect false syllogisms, and deal with other issues such as equivalence relations and false data, then the web of links will transform into the web meaning. References About the Open Directory Project. Retrieved September 22, 2004 from the Open Directory Project website: http://dmoz.org/about.html Berners-Lee, T., & Hendler, J. & Lassila, O. (April 2002). The Semantic Web: A new form of web content that is meaningful to computers will unleash a revolution of new possibilities. Scientific American: Special Online Issue 24-30. Retrieved September 21, 2004 from http://www.kubrussel.ac.be/onderwijs/etew/tew/vakken/tweedekan/ecsector/kennistech/semantic_web.annot.pdf Browser Statistics. Retrieved September 21, 2004 from W3Schools Online Web Tutorials: http://www.w3schools.com/browsers/browsers_stats.asp Chamberlain, Ellen. (2000). Bare Bones: A Basic Tutorial on Searching the Web, Bare Bones Lesson 1: Search Engines. Retrieved September 21, 2004 from http://www.sc.edu/beaufort/library/pages/bones/lesson1.shtml Fensel, D., & Hendler J., & Lieberman, H., & Wahlster, W. (2003). Spinning the semantic web: Bringing the World Wide Web to its full potential. Cambridge, MA: The MIT Press. Retrieved September 22, 2004 from Books 24x7. Flynn, Peter. (2004, May 1). The XML FAQ, B.3 Where can I get an XML browser? Retrieved September 21, 2004 from http://www.ucc.ie/xml/ Guha, P. & McCool, R. & Miller, E. (2003, May), Semantic Search. Proceedings of the twelfth international conference on World Wide Web, 700-709. Abstract retrieved September 22, 2004 from The ACM Digital Library. Ouellet, R. & Ogbuji, U. Introduction to DAML: Part 1. (2002, January 30). Retrieved September 21, 2004 from http://www.xml.com/pub/a/2002/01/30/daml1.html Ontology (computer science). (2004, September 15). Retrieved September 21, 2004 from http://en.wikipedia.org/wiki/Ontology_%28computer_science%29 OWL Web Ontology Language Overview. Retrieved September 22, 2004 from the W3C website: http://www.w3.org/TR/owl-features/ Semantic Web. (2004, September 14). Retrieved September 21, 2004 from http://en.wikipedia.org/wiki/Semantic_web Shirky, Clay. (2003, November 7). The Semantic Web, Syllogism, and Worldview. Retrieved September 22, 2004 from http://www.shirky.com/writings/semantic_syllogism.html Web Service. (2004, September 19). Retrieved September 21, 2004 from http://en.wikipedia.org/wiki/Web_services Welcome to OIL. Retrieved September 21, 2004 from the OIL-page: http://www.ontoknowledge.org/oil/