OntoTag: A Semantic Web Page Linguistic Annotation Model Guadalupe Aguado de Cea1, Inmaculada Álvarez de Mon2, Antonio Pareja-Lora3, Rosario Plaza-Arteche4 Abstract. Although with the Semantic Web initiative much this context, the semantic annotation of texts, since it makes research on web page semantic annotation has already been done meaning explicit, has become a relevant topic and, therefore, by AI researchers, linguistic text annotation, including the advanced design and application of models and formalisms for the semantic one, was originally developed in Corpus Linguistics and semantic annotation of web pages are needed. its results have been somehow neglected by AI. The purpose of Lately, much research has already been carried out by the research presented in this proposal is to prove that integration of results in both fields is not only possible, but also highly useful ontologists on the semantic annotation of web pages [3], [10], [11], in order to make Semantic Web pages more machine-readable. A [12]. However, such works have somehow neglected the results multi-level (possibly multi-purpose and multi-language) obtained on corpus annotation in the field of Corpus Linguistics, annotation model based on EAGLES standards and Ontological not only in the semantic level, but also in other linguistic levels. Semantics, implemented with last generation Semantic Web These other linguistic levels, whilst not being intrinsically languages is being developed to fit the needs of both semantic, can add extra semantic information to help a computer communities.1 2 3 4 understand a text or, in our case, web pages. The goal of this paper is to present the results of our research in which special efforts are being devoted to finding a way of 1. INTRODUCTION. bringing together and identifying complementarities between the All of us are by now used to making extensive use of the so-called semantic annotation models from AI and the annotations proposed World Wide Web (WWW) which we might consider a great source by Corpus Linguistics. of information, accessible through computers but, hitherto, only This paper is organised as follows: firstly, an introduction to understandable to human beings. In its beginning, web pages were the state of the art in semantic annotation in corpus linguistics is hand made, intended and oriented to the exchange of information presented (section 2). In section 3, some brief notes on the use of among human beings. All of these documents contained a huge ontologies in semantic annotation are sketched. In section 4, an amount of text, images and even sounds, meaningless to a example of the integration of both paradigms (AI’s and Corpus computer. In this way, they put the burden of extracting and Linguistics’) is presented in the scope of our project goals. The interpreting the relevant information on the reader. Due to the main advantages of this integration are then analysed –section 5– astonishing growth of Internet use, new technologies emerged and, and, finally, further work to be done is included –section 6–. with them, machine-aided web page generation appeared. Currently, web page presentation in the WWW is being handled independently from its content, mainly through the use of XML [1] 2. SEMANTIC ANNOTATION IN CORPUS or other resource-oriented languages as XOL [2], SHOE [3], OML LINGUISTICS. [4], RDF [5], RDF Schema [6], OIL [7] or DAML+OIL [8]. But The idea of text annotation was originally developed in Corpus even though the automatic process of information is being eased, Linguistics. Traditionally, linguists have defined corpus as "a body still the above-mentioned tasks –relevant information access, of naturally occurring (authentic) language data which can be used extraction and interpretation– cannot be wholly performed by as a basis for linguistic research" [13]. From this point of view, computers. Hence, the goal of enabling computers to understand Corpus Linguistics [14] may not be considered a branch of the meaning (the semantics) of written texts and web pages is the Linguistics in itself, like syntax or semantics. The latter are focused main pillar sustaining the development of the Semantic Web [9]. In on describing or explaining an aspect of language use; the former is –––––––––––––––––––––––––––––––––––––––––––––––––––––– rather a methodology or an approach which can be taken by these 1 Department of Applied Linguistics to Science and Technology (DLACT), branches to explain or describe their particular aspect of language Computer Science Faculty, UPM, Madrid, Spain. lupe@fi.upm.es. use. Following the same authors, Corpus Linguistics was first 2 DLACT, Telecommunications Engineering College, UPM, Madrid, Spain. applied to research on language acquisition, to the teaching of a ialvarez@euitt.upm.es. second language, to the elaboration of descriptive grammars, etc.. 3 Department of Computer Systems and Programming (DSIP), Computer With the arrival of computers, the number of potential studies to Science Faculty, UCM, Madrid, Spain. apareja@sip.ucm.es. which corpora could be applied increased exponentially. 4 Department of Applied Linguistics to Science and Technology (DLACT), Computer Science Faculty, UPM, Madrid, Spain. rplaza@fi.upm.es. So, nowadays, the term corpus is being applied to "a body of decide which one must employ, simply by moving up or down language material which exists in electronic form, and which may to the next level in the hierarchy. be processed by computer for various purposes such as linguistic 6. It should conform to a standard, if one exists. A hard-and-fast research and language engineering" [13]. An annotated corpus system of categories, even being the result of a consensual "may be considered to be a repository of linguistic information [...] work, may be rejected by many researchers. However, a made explicit through concrete annotation" [14]. The benefit of standard in this level could lay, like EAGLES standards have such an annotation is clear: it makes retrieving and analysing done in other levels, a broad framework of principles and information about what is contained in the corpus quicker and major categories. Such a standard would facilitate easier. Let us now see the recommendations stated in Corpus comparability and, at the same time, could be modified as Linguistics for text semantic annotation. necessary for individual needs7. As asserted in [14], two broad types of semantic annotation may be identified, related to: 1. Semantic relationships between items in the text (i.e., the 3. ONTOLOGIES AND SEMANTIC WEB agents or patients of particular actions). This type of annotation ANNOTATIONS. has scarcely begun to be applied. AI researchers have found in ontologies [17], [18] the ideal 2. The semantic features of words in a text, essentially the knowledge model to formally describe web resources and its annotation of word senses in one form or another. There is no vocabulary and, hence, to make explicit in some way the universal agreement in semantics about which features of underlying meaning of the terms included in web pages. With words should be annotated5. Ontological Semantics [19] as a support theory8, the annotation of Although some preliminary recommendations on lexical these web resources with ontological information should allow semantic encoding have already been posited [15], no EAGLES intelligent access to them, should ease searching and browsing semantic corpus annotation standard has yet been published; within them and should exploit new web inference approaches nevertheless, for choosing or devising a corpus semantic field6 from them. Many systems and projects have been developed: annotation system (second type of semantic annotation above SHOE [3]; the (KA)2 initiative [10]; PlanetOnto [11] and the mentioned) a set of reference criteria has been proposed by Semantic Community Web Portals project [12]. Semantic Schmidt and is presented in [16]. These criteria are: annotation tools have also been developed so far: COHSE [20], 1. It should make sense in linguistic or psycholinguistic terms. It MnM [21], OntoMat-Annotizer [22], SHOE Knowledge Annotator is known from psycholinguistic experiments that certain basic [23] and AeroDAML [24]. categories exist in the mind. At present, in general, there is a good agreement between many basic categories we already know about from neuropsychology (for example colours, 4. INTEGRATION OF PARADIGMS: AN body parts, topography and so on); but still an exhaustive set EXAMPLE. of categories is to be determined. Overabstraction must be avoided, in any case. As we have already mentioned, the goal of this paper is to 2. It should be able to account exhaustively for the vocabulary in present the complementarities of linguistic and ontological the corpus, not just for a part of it. If a term cannot readily be annotation for the Semantic Web. The purpose of the project we classified in the existing annotation system, then the system are presenting, ContentWeb, is the creation of an ontology-based clearly needs to be amended. platform to enable users to query e-commerce applications by 3. It should be sufficiently flexible to allow for those using natural language, performing the automatic retrieval of emendations that are necessary for treating a different period, information from web documents annotated with ontological and language, register or textbase. The treatment of specialised linguistic information. ContentWeb objectives can be enunciated as texts (such as computer-related, commerce, etc.) may require follows: considerably more detailed subclassification of the domain in 1. Semi-automatic building of ontologies in the domains of e- question than other texts. commerce and of entertainment, reusing existing ontologies 4. It should operate at an appropriate level of granularity (or and international e-commerce standards and joint initiatives. delicacy of detail) –related to criteria (3). What level of 2. Elaboration of OntoTag, a model and environment for the granularity is correct for an annotation system is an open hybrid –linguistic and ontological– annotation of web question and depends partly on the aims of the end user. For documents. this reason, the next criterion is posited. 3. Development of OntoConsult, a natural language interface 5. It should, where appropriate, possess a hierarchical structure. based on ontologies. If a semantic category system has a hierarchical structure, based on increasingly general levels of relatedness between terms, the end user can look at all the different levels and –––––––––––––––––––––––––––––––––––––––––––––––––––––– 7 Once again the SENSEVAL initiatives [30], [31] must be mentioned: they –––––––––––––––––––––––––––––––––––––––––––––––––––––– reveal the demand for semantic standardization in the field of word sense 5 See, for example, the controversies within the SENSEVAL initiative disambiguation. 8 meetings – [30], [31]. Ontological Semantics [19] uses a constructed world model –the 6 A semantic field (sometimes also called a conceptual field, a semantic ontology– as the central resource for extracting and representing meaning domain or a lexical domain) is a theoretical construct which groups of natural language texts, reasoning about knowledge derived from texts together words that are related by virtue of their being connected –at as well as generating natural language texts based on representations of some level of generality– with the same mental concept [16]. their meaning. be referenced in higher-level relationships or by other levels of the annotation model (i.e. Tras cinco años de espera y después de rdf:ID="1_510">). The annotation of the phrase “la película muchas habladurías, llega a nuestras pantallas la película más esperada de los últimos tiempos” has been included in the más esperada de los últimos tiempos. figure. In the semantic level (see Figure 3) some components of lower level annotations are tagged with semantic references to the concepts, attributes and relationships determined by our (domain) ontology, implemented in the language DAML+OIL. la 5. ADVANTAGES OF THE INTEGRATED MODEL. As shown in the example from section 4, it seems that AI and Corpus Linguistics, far from being irreconcilable, can join together to give birth to an integrated annotation model. This ARTDFS conjunct annotation scheme would be very useful and valuable el in the development of the Semantic Web and would benefit from the results of both disciplines in many ways. Let us now see the benefits at the semantic level of a hybrid annotation TDFS0 model, first from a linguistic point of view and, then, from an el ontological point of view. DET FEM SG la DN> NP los últimos tiempos Figure 1: Morphosyntactic annotation of the article “la”. 4. Creation of OntoAdvice, an ontology-based system PP for querying and retrieving information from de annotated web documents in the entertainment los últimos tiempos domain. One of the tasks performed to reach goal 2 is the manual annotation of a Spanish sentence "Tras cinco años de espera y después de muchas habladurías, llega a nuestras pantallas la película más esperada de los AdjP últimos tiempos." (“After five years of expectation and más gossiping, here comes the most expected film for the esperada time being.”) on the languages XML and RDF(S). The de los últimos tiempos RDF(S) annotation of this sentence in the first three levels is shown in Figure 1, Figure 2 and Figure 3. In the morphosyntactic level (Figure 1) every word or lexical token is given a different Uniform Resource NP Identifier (URI). The morphosyntactic annotation of the la article “la”, according to three different tagsets and película systems is presented. Each tagset has been assigned a más esperada de los últimos different class in the morphAnnot namespace: tiempos TradAnnot (CRATER tagset), MBTAnnot (MBT tagset [25]) and ConstrAnnot (Constraint Grammar - CONEXOR tagset [26]). For the sake of space, just the annotation of the article “la” has been included in the figure. Figure 2: Syntactic annotation of the chunk “la película más In the syntactic level (Figure 2) every syntactic relationship esperada de los últimos tiempos” in RDF(S). between morpho-syntactic items is given a new URI, so that it can llega a nuestras pantallas película después de muchas habladurías Tras cinco años de espera más esperada de los últimos tiempos El Señor de los Anillos El Señor de los Anillos Figure 3: Semantic annotation of "Tras cinco años de espera y después de muchas habladurías, llega a nuestras pantallas la película más esperada de los últimos tiempos." in RDF(S). are more ambitious and potentially wider than the strictly ontology- 5.1. Regarding ontology-based annotations from based ones. Establishing a link between semantic annotation and a linguistic point of view. discourse annotation and text construction following the RST The first result of our work is that the use of ontologies as a approach, which has already been applied in text generation [29], basis for a semantic annotation scheme fits perfectly and seems a fairly promising linguistic enhancement. accomplishes the criteria posited by Schmidt. Clearly, its mostly hierarchical structure fulfils by itself criterion (5) and, as a side effect, criteria (2) and (4), since an ontology can grow horizontally 6. CONCLUSIONS AND FURTHER WORK. (in breadth) and vertically (in depth). Criterion (3) is also satisfied This paper has shown the results of the research carried out on by an ontology-based semantic annotation scheme, since we can how linguistic annotation can help computers understand the text always specialise the concepts in the ontology according to specific contained in a document –a Semantic Web page– bringing together periods, languages, registers and textbases. Ontologies are, by semantic annotation models from AI and the annotations proposed definition, consensual and, thus, are closer to becoming a standard for every linguistic level from Corpus Linguistics. than many other knowledge models, as criteria (6) requires. Further elements susceptible of semantic annotation are Concerning criterion (1), quite a lot of groups developing presently being sought and research is being done towards their ontologies are characterized by a strong interdisciplinary approach determination by the team of linguists in our project. The that combines Computer Science, Linguistics and (sometimes) pragmatic counterpart of OntoTag has not yet been tackled at this Philosophy; then, an ontology-based approach should also make phase of the project. sense in linguistic terms. Still, much work must be done in order to fully specify, implement and assess the whole model. Besides, many efforts are being devoted to developing OntoAdvice, the ontology-based 5.2. Regarding linguistic annotations from an information retrieval system, in order to validate this model. ontological point of view. The main drawback for AI researchers to adopt a linguistically ACKNOWLEDGEMENTS. motivated annotation model would lie on the fact that (section 2) “there is no universal agreement in semantics about which features The research described in this paper is supported by MCyT of words should be annotated” or on Schmidt’s criterion (1): “still (Spanish Ministry of Science and Technology) under the project an exhaustive set of categories is to be determined”. But ontology name: ContentWeb: “PLATAFORMA TECNOLÓGICA PARA researchers are trying to fill this gap with initiatives such as the LA WEB SEMÁNTICA: ONTOLOGÍAS, ANÁLISIS DE UNSPSC [27] or RosettaNet [28] in specific domains (i.e. e- LENGUAJE NATURAL Y COMERCIO ELECTRÓNICO” – commerce). In any case, linguistic annotations at the semantic level TIC2001-2745 ("ContentWeb: Semantic Web Technologic Platform: Ontologies, Natural Language Analysis and E- [18] Studer, R., Benjamins, R., Fensel, D. (1998) Knowledge Business"). We would also like to thank Socorro Bernardos, Óscar Engineering: Principles and Methods. DKE 25(1-2): 161-197. Corcho and Mariano Fernández for their help with the ontological [19] Nirenburg, S. and Raskin, V. (2001) Ontological Semantics (Draft) aspects of this paper. http://crl.nmsu.edu/Staff.pages/Technical/sergei/book/index-book. html. [20] COHSE (2002) http://cohse.semanticweb.org/ REFERENCES. [21] Vargas-Vera, M., Motta, E., Domingue, J., Shum, S. B., Lanzoni, M. [1] Bray, T., Paoli, J., Sperberg, C. (1998) Extensible Markup Language (2001) Knowledge Extraction by Using an Ontology-based (XML) 1.0. W3C Recommendation. http://www.w3.org/TR/REC- Annotation Tool. In Proceedings of the K-CAP'01 Workshop on xml Knowledge Markup and Semantic Annotation, Victoria B.C., [2] Karp, R., Chaudhri, V., Thomere, J. (1999) XOL: An XML-Based Canada. Ontology Exchange Language. Technical Report. [22] OntoMat (2002) http://annotation.semanticweb.org/ontomat.html http://www.ai.sri.com/~pkarp/xol/xol.html [23] SHOE (2002) http://www.cs.umd.edu/projects/plus/SHOE/ [3] Luke S., Heflin J. (2000) SHOE 1.01. Proposed Specification. SHOE KnowledgeAnnotator.html Project. http://www.cs.umd.edu/projects/plus/SHOE/spec1.01.htm [24] AeroDAML (2002) http://ubot.lockheedmartin.com/ubot/hotdaml/ [4] Kent, R. (1998) Conceptual Knowledge Markup Language (version aerodaml.html 0.2). http://sern.ucalgary.ca/KSI/KAW/KAW99/papers/Kent1/ [25] MBT (2002) http://ilk.kub.nl/~zavrel/tagtest.html CKML.pdf [26] Conexor OY (2002) http://www.conexoroy.com/products.htm [5] Lassila, O., Swick, R. (1999) Resource Description Framework [27] UNSPSC (2002) Universal Standard Products and Services (RDF) Model and Syntax Specification. W3C Recommendation. Classification (UNSPSC). http://www.unspsc.org/ http://www.w3.org/TR/PR-rdf-syntax [28] RosettaNet (2002) RosettaNet: Lingua Franca for eBusiness. [6] Brickley, D., Guha, R.V. (2000) Resource Description Framework http://www.rosettanet.org/ (RDF) Schema Specification. W3C Candidate Recommendation. [29] Mann, W., Thomson, S. (1988) Rhetorical Structure Theory: http://www.w3.org/TR/PR-rdf-schema. Toward a functional theory of text organization. Text Vol.18, 3: [7] Horrocks, I., Fensel, D., Harmelen, F., Decker, S., Erdmann, M, 243–281. Klein, M. (2000) OIL in a Nutshell. In 12th International Conference [30] Kilgarriff, A. (1998) SENSEVAL: An Exercise in Evaluating Word in Knowledge Engineering and Knowledge Management, Lecture Sense Disambiguation Programs. In Proceedings of LREC, Granada, Notes in Artificial Intelligence, 1–16. Berlin, Germany: Springer- Spain, pp. 581–588. Verlag. http://www.cs.vu.nl/~ontoknow/oil/downl/oilnutshell.pdf [31] Kilgarriff, A. & Rosenzweig, J. (2000) English SENSEVAL: Report [8] Horrocks, I., Van Harmelen, F. (2001) Reference description of the and Results. In Proceedings of LREC. Athens, Greece. DAML+OIL ontology markup language. Draft report, 2001. http://www.daml.org/2000/12/reference.html [9] Berners-Lee, T., Fischetti, M. (1999) Weaving the Web: The Original Design and Ultimate Destiny of the World Wide Web by its Inventor. San Francisco: Harper. [10] Benjamins, V.R., Fensel, D., Decker, S., Gómez-Pérez, A. (1999) (KA)2: Building Ontologies for the Internet: a Mid Term Report. IJHCS, International Journal of Human Computer Studies, 51: 687– 712. [11] Motta, E., Buckingham Shum, S. Domingue, J. (1999) Case Studies in Ontology-Driven Document Enrichment. In Proceedings of the 12th Banff Knowledge Acquisition Workshop, Banff, Alberta, Canada. [12] Staab, S., Angele, J., Decker, S., Erdmann, M., Hotho, A., Mädche, A., Schnurr, H.-P., Studer, R. (2000) Semantic Community Web Portals. WWW´9. Amsterdam. [13] Leech, G. (1997a) Introducing corpus annotation. In Garside R., Leech, G., McEnery, A. M. (eds.) Corpus Annotation: Linguistic Information from Computer Text Corpora. London: Longman. [14] McEnery, A. M., Wilson, A. (2001) Corpus Linguistics: An Introduction. Edinburgh: Edinburgh University Press. [15] EAGLES (1999) EAGLES LE3-4244: Preliminary Recommendations on Semantic Encoding, Final Report. http://www.ilc.pi.cnr.it/EAGLES/EAGLESLE.PDF [16] Wilson, A., Thomas, J. (1997) Semantic Annotation. In R. Garside, G. Leech & A. M. McEnery, (eds.) Corpus Annotation: Linguistic Information from Computer Text Corpora. London: Longman. [17] Gruber, R. (1993) A translation approach to portable ontology specification. Knowledge Acquisition. #5: 199-220.