Fuzzy Conceptual Knowledge Processing
- Christoph S. Herrmann,Steffen Holldobler,Antje Strohmaier
We introduce fuzziness to conceptual knowledge processing
by using linguistic variables instead of a two-valued
representation. The attribute/object table for conceptual
lattices holds fuzzy membership values rather than
TRUE/FALSE entries and can be mapped into a graph
of dependencies. From this graph implications can be extracted
together with the method to compute truth values
for the inferible conclusions. Hence, fuzzy conclusions can
be drawn from interpreting the fuzzy concept values of the
graph. We demonstrate the feasibility of our approach by
computing patient data from a medical diagnosis example.
The analysis of large data sets can be achieved in a way
that is closer to human knowledge representation than
simple lists of...
A Feature-Relevance Heuristic for Indexing and Compressing Large Case Bases
- Walter Daelemans,Antal Van Den Bosch,Jakub Zavrel
. This paper reports results with igtree, a formalism for indexing
and compressing large case bases in Instance-Based Learning (ibl)
and other lazy-learning techniques. The concept of information gain (entropy
minimisation) is used as a heuristic feature-relevance function for
performing the compression of the case base into a tree. igtree reduces
storage requirements and the time required to compute classifications
considerably for problems where current ibl approaches fail for complexity
reasons. Moreover, generalisation accuracy is often similar, for
the tasks studied, to that obtained with information-gain-weighted variants
of lazy learning, and alternative approaches such as c4.5. Although
igtree was designed for a specific class of problems --linguistic disambiguation
problems with symbolic...
The Role of Morphology in Machine Translation
- Bowen Hui
this paper: to apply various aspects of
morphology to MT.
The outline of this paper follows. Section 2 provides an overview of three linguistic approaches in
machine translation. These methods are direct, transfer, and interlingua. A summary with a comparison
of the three approaches is given. We observe that many problems arise from lack of morphological theories,
which leads to the discussions in section 3 and 4. Section 3 briefly introduces various aspects of morphology
in linguistics. It reviews different types of word formation processes, such as derivation, inflection,
compounding, clipping, and blending, as well as motivating the need for a two-level morphological theory
for accurate performance. Section...
-- A Unified Platform for Parsing and Generation
- Shuly Wintner,Evgeniy Gabrilovich,Nissim Francez
Contemporary linguistic theories (in particular,
HPSG) are declarative in nature: they specify
constraints on permissible structures, not how
such structures are to be computed. Grammars
designed under such theories are, therefore,
suitable for both parsing and generation.
However, practical implementations of such theories
don't usually support bidirectional processing
of grammars. We present a grammar
development system that includes a compiler of
grammars (for parsing and generation) to abstract
machine instructions, and an interpreter
for the abstract machine language. The generation
compiler inverts input grammars (designed
for parsing) to a form more suitable for generation.
The compiled grammars are then executed
by the interpreter using one control strategy, regardless
of whether the grammar is the original
or the inverted...
Semantic Hidden Markov Networks
- G. A. Fink,F. Kummert,G. Sagerer,E. G. Schukat-talamazzini,H. Niemann
Although much effort has been put into speech understanding systems
there still exists a rather wide gap between acoustic recognition
and linguistic interpretation. We propose a formalism for an
extremely close interaction of acoustic recognition and higher level
analysis. Instead of a strict horizontal interface at the level of hypothesized
word sequences or lattices, a vertical interface to the acoustic
component is used that can be accessed from linguistic concepts of
any degree of abstraction. As the linguistic knowledge is represented
in the formalism of Semantic Networks and acoustic recognition is
based on Hidden Markov Models the close interaction between the
two components was termed Semantic Hidden Markov Networks.
Intelligent Agents with Subjective Experience
- Ana Pasztor
In this paper I am presenting a model of human communication
which allows a decomposition of our subjective experience
into units called strategies. Strategies are systematically
ordered sequencesof representations in our sensory modalities,
endowed with so called submodality distinctions, orientation,
effect, and links to other representations. As certain people
prove more successful in performing given tasks than others,
we want to model their strategies and transfer them to other
intelligent agents. Unlike most philosophers who have written
about qualia and believe that subjective experience condemns
us to "solitary confinement," I believe strategies are accessible
to us. Here is how. All behavior, from learning to communication
and change, is the result of our...
Programming with Structures, Functions, and Objects
- Fritz Henglein,Konstantin Laufer
We describe program structuring mechanisms for integrating algebraic, functional and objectoriented
programming in a single framework. Our language is a statically typed higher-order
language with specifications, structures, types, and values, and with universal and existential
abstraction over structures, types, and values.
We show that existential types over structures generalize both the necessarily homogeneous
type classes of Haskell and the necessarily heterogeneous object classes of object-oriented programming
languages such as C++ or Eiffel. Following recent work on ML, we provide separate
linguistic mechanisms for reusing specifications and structures. Subtyping is provided in the
form of explicit type conversions.
The language mechanisms are introduced by examples to emphasize their pragmatic aspects.
SRI International, 23 Millers Yard, Cambridge CB2 1RQ United Kingdom
- Ian Lewin,Ralph Becket,Johan Boye,David Carter,Manny Rayner
With maturing speech technology, spoken dialogue
systems are increasingly moving from research prototypes
to fielded systems. The fielded systems however
generally employ much simpler linguistic and dialogue
processing strategies than the research prototypes.
We describe an implemented spoken-language
dialogue system for a travel planning domain which
supports a mixed initiative dialogue strategy. The system
accesses a commercially available travel information
web-server. The system architecture combines
both shallow and deep linguistic processors, partly so
that a robust if shallow analysis is always available to
the dialogue manager, and partly so that we can begin
to examine where significant gains can be made by
employing more advanced linguistic processing. We
present the results of a preliminary investigation...
Lexicalization of Context
- Graham Wilcock
We propose lexical rather than phrasal amalgamation of contextual features. On a theoretical level, the lexicalization of context naturally follows other recent revisions concerned with lexicalization of set-valued features (nonlocal features, quantifier storage). On a computational level, it offers advantages in using HPSG with head-driven generation algorithms. On a linguistic level, we sketch a lexical context approach to register variation in English relative clauses.
Test Environment for the Two Level Model of Germanic Prominence
In this work we present a test bed designed to verify
the two level model of Germanic prominence. We
give an introduction to the linguistic background of
the model and derive the features that the test environment
should possess. Finally we describe the
details of the implementation in the ESPS/xwaves
environment. The implementation is based on
resynthesis using PSOLA algorithm. As a linguistic
application the Tone Sequence Model (TSM) is
implemented and tested.
Keywords: Prosody, Modelling of F0, PSOLA,
Recent developments in the theory of prominence
 rigorously restrict the number of categories
upon which prominence relations may be expressed.
Of the number of prosodic constituents proposed in
the early 80s (see ...
A Note on the Complexity of Restricted Attribute-Value Grammars
- Leen Torenvliet,Marten Trautwein
The recognition problem for attribute-value grammars(AVGs) was
shown to be undecidable by Johnson in 1988. Therefore, the general
form of AVGs is of no practical use. In this paper we study a very
restricted form of AVG, for which the recognition problem is decidable
(though still NP-complete), the R-AVG. We show that the R-AVG
formalism captures all of the context free languages and more, and
introduce a variation on the so-called off-line parsability constraint ,
the honest parsability constraint , which lets different types of R-AVG
coincide precisely with well-known time complexity classes.
Although a universal feature theory does not exist, there is a general understanding
of its objects....
History Integration Into Semantic Classification
- M. Cettolo,A. Corazza
In spoken language systems, the classification of coherent
linguistic/semantic phrases in terms of semantic classes
is an important part of the whole understanding process.
Basically, it relies on the plain text of the segment to
be classified. Nevertheless, another important source of
useful information is the dialogue context. In this paper, a
number of different ways to integrate the dialogue history
into the semantic classification are presented and tested
on a corpus of person-to-person dialogues. Best result
gives a 3.6% reduction of the error rate with respect to
the performance obtained without using history.
In machine-mediated person-to-person communication,
care should be devoted to preserve communication in a
robust way. The task considered...
Presuppositions in Context: Constructing Bridges
- Paul Piwek,Emiel Krahmer
this article, we want to get a formal grip on the way
in which context influences the behaviour of presuppositions. Before we describe
how we intend to do this, let us first describe the notion of context we are interested
in. There are various uses of the term `context'. Bunt (1995) characterizes context
as all those factors which are relevant to the understanding of communicative
behaviour, and he goes on to distinguish five major dimensions: the linguistic context,
the semantic context, the physical context, the social context and the cognitive
context. For presuppositions in general, and for bridging in particular, the following
seem most relevant: the linguistic context,...
What Have the Breasts of a Running Woman Got to Do With Relevance?
- Katalin Egri Ku-mesu
The aim of this paper is to investigate, through the analysis of interpretations
of African proverbs, how different audiences in different sociolinguistic contexts may
interpret `hybrid texts', i.e. texts written by authors of a particular cultural and
linguistic background in a language of a different culture, which, therefore, raise the
question of how cultural experience is, or can be, communicated in an `alien' language.
The data were collected from twenty subjects by means of a questionnaire.
In hybrid literary texts, along with riddles, dirges, praise names and different forms
of address, one relatively often encounters proverbs. These constitute what Zabus calls
the ethno-text, `which is grafted on the...
Towards a Systematic Evaluation of Machine Translation Systems
- Melchior Amgarten,Dominic A. Merz
There are still a lot of uncertainties about the evaluation of machine translation
(MT) systems. In this report, we try to establish a catalog of criteria which we consider
to be crucial for a systematic assessment of MT systems; the paper is divided
into three parts: technical, linguistic and entrepreneurial criteria. Furthermore, we
give hints on how to organize an evaluation project.
It is not easy to decide on the right criteria for evaluating machine translation (MT)
systems (cf. [Kin96]), but it is even more difficult to find precise measures which objectively
describe what a good system consists of, or to find a suitable evaluation setup.
Computing Spatio-Temporal Representations of Human Faces
- Yaser Yacoob,Larry Davis
An approach for analysis and representation of
facial dynamics for recognition of facial expressions
from image sequences is proposed. The algorithms
we develop utilize optical flow computation to identify
the direction of rigid and non-rigid motions that are
caused by human facial expressions. A mid-level symbolic
representation that is motivated by linguistic and
psychological considerations is developed. Recognition
of six facial expressions, as well as eye blinking, on a
large set of image sequences is reported.
Human visual communication has been extensively
studied in the social and psychology literature, mainly
as a means to describe the emotional state of the subject
[3,5,11,15]. Research in psychology has indicated
that at least six emotions are...
Measuring the Complexity of Writing Systems
- Walter Daelemans,Beatrice De Gelder
this paper, we take the latter approach
and disregard the complexity of converting
phonemic representations to spelling.
Based on the (disputed) view that reading aloud
involves two independent processes, viz. direct
word pronunciation using lexical retrieval, and rulebased
grapheme-to-phoneme conversion (i.e., the
dual-route model, Coltheart, 1978), cross-linguistic
experiments seem to indicate that the balance between
these two processes varies as a function of the
orthographic depth of the language. More specifically,
several authors claim that in shallow orthographies,
such as Serbo-Croatian, the analytic
rule-based route, operating on grapheme--phoneme
correspondences (GPCs), is used more intensively
than the lexical retrieval route (cf. Frost et al.,
1987). The rationale behind this claim is that using
the GPC-based route in a...
Interpreting Changes In The Fuzzy Sets Of A Self-Adaptive Neural Fuzzy Controller
- Detlef Nauck,Rudolf Kruse
We describe a procedure for the adaptation of membership functions in a fuzzy control
environment by using neural network learning principles. The changes in the fuzzy sets
can be easily interpreted. By using a fuzzy error that is propagated back through the
architecture of our fuzzy controller, we receive an unsupervised learning technique, where
each rule tunes the membership functions of its antecedent and its consequence.
Classical control theory is based on mathematical models that describe the behaviour of the plant under
consideration. The main idea of fuzzy control [9, 10], which has proved to be a very successful method ,
is to build a model of...
Elementary Principles of HPSG
- Georgia M. Green
This chapter describes the theoretical foundations and descriptive mechanisms of Head-Driven
Phrase Structure Grammar (HPSG), as well as proposed treatments for a number of familiar
grammatical phenomena. The anticipated reader has some familiarity with syntactic phenomena
and the function of a theory of syntax, but not necessarily any expertise with modern theories of
phrase-structure grammar. Section 1 describes the character of HPSG grammars, and the elements
and axioms of the system. Section 2 describes how linguistic entities are modelled, and how
grammars describe the modelled entities. The third section describes the structure of featurestructure
descriptions in HPSG, and Section 4 deals with the expression of constraints, especially
Syriac Morphology: From a Linguistic Model to a Computational Implementation
- George Anton Kiraz,Syriac Grammar
This paper proposes a linguistically-motivated model of Syriac morphology which
can be implemented in a computational environment. My aim is to formalise the
derivation of Syriac stems based on the data provided in Noldeke (1904: x92 ff., x160