No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Pros and Cons of the Pivot and Transfer Approaches in Multilingual Machine Translation
... Consider another example from the language domain, with tasks given by: As above, if the learner has learned to solve tasks Z (1) and Z (2) , it could solve task Z (3) by first translating the text from English to Spanish and subsequently translating the resulting text from Spanish to Italian. Here, Spanish would act as a pivot language (Boitet, 1988). This definition can be applied to RL problems as well. ...
A major goal of artificial intelligence (AI) is to create an agent capable of acquiring a general understanding of the world. Such an agent would require the ability to continually accumulate and build upon its knowledge as it encounters new experiences. Lifelong or continual learning addresses this setting, whereby an agent faces a continual stream of problems and must strive to capture the knowledge necessary for solving each new task it encounters. If the agent is capable of accumulating knowledge in some form of compositional representation, it could then selectively reuse and combine relevant pieces of knowledge to construct novel solutions. Despite the intuitive appeal of this simple idea, the literatures on lifelong learning and compositional learning have proceeded largely separately. In an effort to promote developments that bridge between the two fields, this article surveys their respective research landscapes and discusses existing and future connections between them.
... However, there have been a number of arguments for the more commonly used transfer approach as an alternative to the interlingual approach. (See, e.g., Arnold and Sadler (1990), Boitet (1988), and Vauquois and Boitet (1985).) Paradoxically, these anti-interlingual arguments are based precisely on the same types of examples that have motivated the current research (e.g., those sentences that exhibit the types of divergences shown above). ...
This paper describes the lexical-semantic basis for UNITRAN, an implemented scheme for translating Spanish, English, and German bidirectionally. Two claims made here are that the current representation handles many distinctions (or divergences) across languages without recourse to language-specific rules and that the lexical-semantic framework provides the basis for a systematic mapping between the interlingua and the syntactic structure. The representation adopted is an extended version of lexical conceptual structure which is suitable to the task of translating between divergent structures for two reasons: (1) it provides an abstraction of language-independent properties from structural idiosyncrasies; and (2) it is compositional in nature. The lexical-semantic approach addresses the divergence problem by using a linguistically grounded mapping that has access to parameter settings in the lexicon. We will examine a number of relevant issues including the problem of defini...
Mit der vorliegenden Bibliographie wird keinesfalls angestrebt, die wissenschaftliche Literatur zur Maschinellen Übersetzung vollständig zu erfassen, denn für verschiedene Forschungszeiträume stehen bereits umfangreiche Bibliographien zur Verfügung1. Hinzu kommt, daß der rasante Anstieg v.a. der sogenannten grauen Literatur in diesem Bereich eine Erfassung durch eine Einzelperson ohne entsprechende finanzielle Ressourcen unmöglich macht. So ist es für Studenten der Übersetzungswissenschaft, der Computerlinguistik oder angrenzender Philologien außerordentlich schwierig, sich in ein bestimmtes Teilgebiet einzuarbeiten. Diese Bibliographie soll daher Hilfestellungen für eine grundlegende Beschäftigung leisten. Wünschenswert wären aus Benutzersicht sicherlich noch eine Kommentierung und ein Register gewesen. Doch der dazu nötige Zeitaufwand konnte von mir leider nicht geleistet werden.
Machine translation (MT) was one of the first non-numerical applications of the computer in the 1950s and 1960s. With limited equipment and programming tools, researchers from a wide range of disciplines (electronics, linguistics, mathematics, engineering, etc.) tackled the unknown problems of language analysis and processing, investigated original and innovative methods and techniques, and laid the foundations not just of current MT systems and computerized tools for translators but also of natural language processing in general. This volume contains contributions by or about the major MT pioneers from the United States, Russia, East and West Europe, and Japan, with recollections of personal experiences, colleagues and rivals, the political and institutional background, the successes and disappointments, and above all the challenges and excitement of a new field with great practical importance. Each article includes a personal bibliography, and the editor provides an overview, chronology and list of sources for the period.
Dialogue-Based Machine Translation (DBMT) is a new paradigm for translation situations where other approaches, such as the Linguistic-Based (LBMT) and the Knowledge-Based (KBMT) approaches, are not adequate. In DBMT, although the linguistic knowledge sources are still crucial, and extralinguistic knowledge might be used if available, emphasis is on indirect pre-editing through a negotiation and a clarification dialogue with the author in order to get high quality translations without revision. Authors are distinguished from "spontaneous" writers or speakers by the fact that they want to produce a "clean" final message and may be willing to enter into such dialogues. After having described the main situational, linguistic and ergonomic issues in DBMT for monolingual authors, we describe ongoing work on the LIDIA project. The typical translational situation considered is the production of multilingual technical documentation in the form of HyperCard stacks. Notable points in the linguistic design include multilevel transfer with interlingual acceptions, properties and relations, the "guided language" approach (typed textual fragments and lexical preferences), and a TEI-inspired representation of texts and structures. The current mockup, LIDIA-1.0, demonstrates the majority of these ideas on a HyperCard stack, to be translated from French into German, Russian and English. Some of its aspects are discussed in detail, in particular the user interface, the object-oriented implementation, and the production of disambiguation dialogues.
ResearchGate has not been able to resolve any references for this publication.