5 <title>Project Objectives</title>
8 <h1>Project Objectives</h1>
9 <p>The new frontier of Content Based Information Systems is the so called
11 <a href="publications/others/w3c_bl98.html">others/w3c_bl98</a>).
12 Associating meaning with content or establishing a layer of machine
13 understandable data will allow automated agents, sophisticated search
14 engines and interoperable services and will enable higher degree
15 of automation and more intelligent applications. The ultimate goal of the
16 Semantic Web is to allow machines to share and exploit knowledge in the
17 Web way, i.e. without central authority, with few basic rules, in a
18 scalable, adaptable, extensible manner. However, the actual development
19 of the Semantic Web and its technologies has been hindered so far by the
20 lack of large scale, distributed repositories of structured, content
21 oriented information. The case of Mathematical knowledge, the most
22 rigorous and condensed form of knowledge, is paradigmatic. The World Wide
23 Web is already now the largest single resource of mathematical knowledge,
24 and its importance will be exponentiated by the emerging display
25 technologies like MathML. However, almost all mathematical documents
26 available on the Web are marked up only for presentation (in this respect,
27 current practice in MathML improves on, but does not fundamentally differ
28 from the older paper-oriented markup schemes like {\LaTeX} or Postscript).
29 A consequence of this is that the online material is machine-readable, but
30 not machine-understandable, severely crippling the possibility to offer
31 added-value services like</p>
33 <li>Preservation of the real informative content in a highly structured and
34 machine understandable format, suitable for transformation, automatic
35 elaboration and processing.</li>
36 <li>Cut and paste on the level of computation (take the output from a Web
37 search engine and paste it into a computer algebra system).</li>
38 <li>Automatic proof checking of published proofs.</li>
39 <li>Semantical search for mathematical concepts (rather than keywords).</li>
40 <li>Classification: given a concrete mathematical structure, is there a
41 general theory for it?</li>
43 <p>Due to its rich notational, logical and semantical structure, mathematical
44 knowledge is thus a main case study for the development of the new
45 generation of semantic Web systems. The aim of the proposed project is
46 both to help in this process, as well as pave the way towards a really
47 useful virtual, distributed, hyper-textual resource for the working
48 mathematician, scientist or engineer. All modern sciences have a
49 strongly mathematicised core, and will benefit. The real market and
50 application area for the techniques developed in this project, apart from
51 the obvious realm of education, lies with high-tech and engineering
52 corporations that rely on huge formula databases. Currently, both the
53 content markup as well as the added-value services alluded to above are
54 very underdeveloped, limiting the usefulness of the vital knowledge. The
55 infrastructure and knowhow needed for mining this information treasure
56 and obtaining a competitive edge in development is exactly what we are
57 attempting to develop in our project.</p>
58 <p>Several languages have been already proposed for the management of
59 mathematical information on the Web, both for publishing, communication
60 and archiving purposes: most notably,
61 <a href="http://www.w3.org/TR/MathML2/">MathML</a>,
62 <a href="http://www.nag.co.uk/projects/openmath/omsoc/">OpenMath</a>,
63 <a href="http://www.mathweb.org/omdoc/">OMDoc</a>. Other languages
64 must be also considered for definition and specification of Metadata,
65 such as the <a href="http://purl.org/dc/">Dublin Core</a> System, or
66 the Resource Description Framework
67 [<a href="http://www.w3.org/RDF/">RDF</a>].
68 All these languages, which tend to cover different and orthogonal aspects
69 of the management of mathematical documents, must be eventually taken into
70 account for the ambitious goal of our project. One of our aims is actually
71 the definition of a modular architecture which could exploit the
72 distinctive potentialities of each one of these languages, integrating
73 them into a single application. The integration is in this case
74 facilitated by the fact that all the languages mentioned are particular
75 instances of XML, providing the opportunity to use standard XML
76 technology, and in particular XSL Transformations or
77 stylesheets [<a href="http://www.w3.org/TR/xslt">XSLT</a>], to pass from
78 one language to the other.</p>
80 <img border="0" alt="Architecture" src="./../images/arch.gif" />
82 <p>The fact of encoding also the microscopic, logical level of mathematics
83 opens the possibility to have completely formalised subsystems of the
84 library, which could be checked automatically by standard tools for the
85 automation of formal reasoning and the mechanisation of mathematics
86 (proof assistants and logical frameworks, see
87 <a href="publications/others/cup_hp91.html">others/cup_hp91</a> and
88 <a href="publications/others/cup_hp93.html">others/cup_hp93</a>). At
89 the same time, any of these tools could be used as an authoring system for
90 documents of the library, by simply exporting their internal libraries
91 into XML, and using stylesheets to transform the output into a standard,
92 machine-understandable representation, such as MathML content markup or
94 <p>The precise formal content can still be preserved by the machinery of
95 <a href="http://www.w3.org/TR/xlink/">Xlinks</a>. Moreover, stylesheets
96 can be also used to solve the annoying notational problem that usually
97 afflicts formal mathematics, providing a simple way for adding
98 user-defined styles and notations.</p>
100 <p>So, our approach leads to a natural integration of proof assistant tools
101 and the Web. In this integration, the emphasis is just on ``content'':
102 we do not try to link directly the specific applications to the Web,
103 that would be a major mistake, for obvious modularity reasons. On the
104 contrary, we adopt XML as a neutral specification language, and then we
105 merely work on XML-documents, forgetting the underlying application. In
106 this way, similar software tools can be applied to different logical
107 dialects, regardless of their concrete nature. Moreover, if having a
108 common representation layer is not the ultimate solution to all
109 inter-operability problems between different applications, it is
110 however a first and essential step in this direction. Finally, this
111 ``standardisation'' process should naturally lead to a substantial
112 simplification and re-organisation of the current, ``monolithic''
113 architecture of logical frameworks. All the many different and often
114 loosely connected functionalities of these complex programs (proof
115 checking, editing, search and consulting, program extraction, and so on)
116 could be clearly split in more or less autonomous tasks, and could be
117 developed by different teams, in totally different languages. This is
118 the new, ``content-based'' architectural design of future systems.</p>