New Trends of Research in Ontologies and Lexical Resources

‘Open Data’ has become very important in a wide range of fields. However for linguistics, much data is still published in proprietary, closed formats and is not made available on the web. We propose the use of linked data principles to enable language resources to be published and interlinked openly on the web, and we describe the application of this paradigm to the modeling of two resources, WordNet and the MASC corpus. Here, WordNet and the MASC corpus serve as representative examples for two major classes of linguistic resources, lexical-semantic resources and annotated corpora, respectively.Furthermore, we argue that modeling and publishing language resources as linked data offers crucial advantages as compared to existing formalisms. In particular, it is explained how this can enhance the interoperability and the integration of linguistic resources. Further benefits of this approach include unambiguous identifiability of elements of linguistic description, the creation of dynamic, but unambiguous links between different resources, the possibility to query across distributed resources, and the availability of a mature technological infrastructure. Finally, recent community activities are described.

Linguistic description is important for the re-use of lexical resources and the interpretation of text. Linguistic knowledge plays an important role in defining and enriching ontological knowledge. The fact that there are multiple proposed (de facto) standard models from the terminological, linguistic and localization fields creates a need for interoperability between linguistic models. These models complement each other or overlap to a certain extent, which creates linguistic confusion. This paper presents the LingNet model and its implementation, which enables interoperability between and comparison of different models, and the harmonization of linguistic description across application domains, allowing a user to a customized combination of elements from different models according to criteria of coverage, complementarity and granularity of linguistic description.

This chapter investigates the notion of ‘sense’ in the ontology-lexicon interface.As a realization of the ontology-lexicon interface, we are concerned with so called ‘ontology lexica’ which specify the meaning of lexical entries by reference to a given ontology.We propose that in the context of the ontology-lexicon interface a ‘sense’ can be understood as a three-faceted entity, i.e.as a (i) reification of the link between a lexical entry and the ontological reference, (ii) as subset of all the uses of the word that can be interpreted as referring to the same ontological reference, and (iii) as an implicitly defined subconcept.We also provide a new definition of the traditional notions of homonymy, synonymy, metonymy etc.in the ontology-lexicon interface.

The recent achievements in Natural Language Processing in terms of scalability and performance, and the large availability of background knowledge within the Semantic Web and the Linked Open Data initiative, encourage researchers in doing a further step towards the creation of machines capable of understanding multimedia documents by exploiting background knowledge. To pursue this direction it turns out to be necessary to maintain a clear link between knowledge and the documents containing it. This is achieved in the KnowledgeStore, a scalable content management system that supports the tight integration and storage of multimedia resources and background and extracted knowledge. Integration is done by (i)identifying mentions of named entities in multimedia resources, (ii)establishing mention coreference and either (iii)linking mentions to entities in the background knowledge, or (iv)extending that knowledge with new entities. We present the KnowledgeStore and describe its use in creating a large scale repository of knowledge and multimedia resources in the Italian Trentino region, whose interlinking allows us to explore advanced tasks such as entity-based search and semantic enrichment.

In the last decade, the annotation of multimedia has evolved toward the use of ontologies, as a way to bridge the semantic gap between low level features of media objects and high level concepts. In many cases, the annotation terms refer to structured ontologies. Such ontologies, however, are often light scale domain oriented knowledge bases, whereas the employment of wide, commonsense ontologies would improve interoperability and knowledge sharing, with beneficial effects on search and navigation. In this chapter, we present an approach to the semantic annotation of media objects through a meaning negotiation approach that requires natural language lexical terms as interface and employs large scale commonsense ontologies. As a test case, we apply the annotation to narrative media objects, using a meta–ontology, called Drammar, to describe their structure. We present the annotation schema, the software architecture for integrating several large scale ontologies, and the lexical interface for negotiating the ontological term. We also describe an evaluation of the proposed approach, conducted through experiments with annotators.

In this work we present the Cognitive Engine, an integrated system whose architectural characteristics and operational capabilities are designed to approximate human visual intelligence. As humans usually do, the Cognitive Engine tries to make sense of a scene by meaningfully clustering visual data: basic individual movements are interpreted as constituting a particular action, and patterns of actions are gathered into more complex activities. In this respect, the Cognitive Engine results from augmenting the ACT-R cognitive architecture – a modular computational system used to model human cognitive processes – with relevant background knowledge embedded in HOMinE, a semantic resource for actions.

This chapter presents an implemented algorithm for resolving reference within the theory of Ontological Semantics with an emphasis on the use of static knowledge resources: ontology—a world model of entity types; fact repository—a world model of entity tokens; and lexicon, which mediates between language and the ontology and fact repository. We show how reference resolution is tightly coupled with overall semantic analysis, from the first stages of determining which expressions have referential function to the final stage of creating a reference link from each referring expression in a text to its “anchor” in the model of memory of the intelligent agent processing the text. As such, there is no single reference resolution task; rather, reference-related subtasks are best distributed throughout an end-to-end text analysis system.

We present an ontology-based semantic interpreter that can be linked to a grammar through grammar rule constraints, providing access to meaning during language processing. In this approach, the parser will take as input natural language utterances and will produce ontology-based semantic representations. We rely on a recently developed constraint-based grammar formalism, which balances expressiveness with practical learnability results. We show that even with a lightweight ontology, the semantic interpreter at the grammar rule level can help remove erroneous parses obtained when we do not have access to meaning.

The exponential growth of available electronic data is almost useless without efficient tools to retrieve the right information at the right time. This is especially crucial in the context of decision making (e.g. for politicians), innovative development (e.g. for scientists and industrials) or economic development (e.g. for market or concurrence studies). It is now widely acknowledged that information retrieval systems (IRS in short) need to take semantics into account. In this context, semantic Web technologies have been rapidly widespread and accepted. This article surveys semantic based methodologies designed to efficiently retrieve and exploit information. Some of them, based on terminologies, are fitted to open context, dealing with heterogeneous and unstructured data, while others, based on taxonomies or ontologies, are semantically richer but require formal knowledge representation of the studied domain. Hence, a continuum of solutions exists from terminology to ontology based IRSs. These approaches are often seen as concurrent and exclusive, but this chapter asserts that their advantages may be efficiently combined in a hybrid solution built upon domain ontology. The original approach presented here benefits from both lexical and ontological document description, and combines them in a software architecture dedicated to information retrieval in specific domains. Relevant documents are first identified via their conceptual indexing based on domain ontology, and then each document is segmented to highlight text fragments that deal with users’ information needs.The system thus specifies why these documents have been chosen and facilitates end-user information gathering.

In the past years, there has been a growing interest in developing computational methods for affect detection from text. Although much research has been done in the field, this task still remains far from being solved, as the presence of affect is only in a very small number of cases marked by the presence of emotion-related words. In the rest of the cases, no such lexical clues of emotion are present in text and special commonsense knowledge is necessary in order to interpret the meaning of the situation described and understand its affective connotations. In the light of the challenges posed by the detection of emotions from contexts in which no lexical clue is present, we proposed and implemented a knowledge base – EmotiNet – that stores situations in which specific emotions are felt, represented as “action chains”. Following the initial evaluations, in this chapter, we describe and evaluate two different methods to extend the knowledge contained in EmotiNet: using lexical and ontological knowledge. Results show that such types of knowledge sources are complementary and can help to improve both the precision, as well as the recall of implicit emotion detection systems based on commonsense knowledge.

Our affective perspective on a word is heavily influenced by the context in which it is used and by the features it is typically perceived to exhibit in that context. A nuanced model of lexical affect thus requires a feature-rich representation of each word’s potential to mean different things in different contexts. To this end, we present here a two-level model of lexical affect. At the first level, words are represented as bundles of the typical properties and behaviors they are commonly shown to exhibit in everyday language. To construct these bundles, we present a semi-automatic approach to harvesting stereotypical properties and behaviors from the Web. At the second level, these properties and behaviors are related to each other in a graph structure that captures how likely one is to reinforce the meaning of another. We present an effective means of constructing such a graph from a combination of text n-grams and queries to the open Web. We calculate positive and negative potentials for each property in the graph, and show how these potentials can be used in turn to calculate an overall affective value for the higher-level terms for which they are considered stereotypical.