BS ISO 24617-6-2016 Language resource management Semantic annotation framework Principles of semantic annotation (SemAF Principles)《语言资源管理 语义注释框架 语义标注原则 (SemAF原则)》.pdf

《BS ISO 24617-6-2016 Language resource management Semantic annotation framework Principles of semantic annotation (SemAF Principles)《语言资源管理 语义注释框架 语义标注原则 (SemAF原则)》.pdf》由会员分享，可在线阅读，更多相关《BS ISO 24617-6-2016 Language resource management Semantic annotation framework Principles of semantic annotation (SemAF Principles)《语言资源管理 语义注释框架 语义标注原则 (SemAF原则)》.pdf（40页珍藏版）》请在麦多课文档分享上搜索。

1、BSI Standards PublicationBS ISO 24617-6:2016Language resourcemanagement Semanticannotation frameworkPart 6: Principles of semantic annotation(SemAF Principles)BS ISO 24617-6:2016 BRITISH STANDARDNational forewordThis British Standard is the UK implementation of ISO 24617-6:2016.The UK participation

2、in its preparation was entrusted to TechnicalCommittee TS/1, Terminology.A list of organizations represented on this committee can beobtained on request to its secretary.This publication does not purport to include all the necessaryprovisions of a contract. Users are responsible for its correctappli

3、cation. The British Standards Institution 2016.Published by BSI Standards Limited 2016ISBN 978 0 580 87548 9ICS 01.020Compliance with a British Standard cannot confer immunity fromlegal obligations.This British Standard was published under the authority of theStandards Policy and Strategy Committee

4、on 29 February 2016.Amendments/corrigenda issued since publicationDate T e x t a f f e c t e dBS ISO 24617-6:2016 ISO 2016Language resource management Semantic annotation framework Part 6: Principles of semantic annotation (SemAF Principles)Gestion des ressources linguistiques Cadre dannotation sman

5、tique Partie 6: Principes dannotation smantique (SemAF Principes)INTERNATIONAL STANDARDISO24617-6First edition2016-02-01Reference numberISO 24617-6:2016(E)BS ISO 24617-6:2016ISO 24617-6:2016(E)ii ISO 2016 All rights reservedCOPYRIGHT PROTECTED DOCUMENT ISO 2016, Published in SwitzerlandAll rights re

6、served. Unless otherwise specified, no part of this publication may be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on the internet or an intranet, without prior written permission. Permission can be requested from either

7、ISO at the address below or ISOs member body in the country of the requester.ISO copyright officeCh. de Blandonnet 8 CP 401CH-1214 Vernier, Geneva, SwitzerlandTel. +41 22 749 01 11Fax +41 22 749 09 47copyrightiso.orgwww.iso.orgBS ISO 24617-6:2016ISO 24617-6:2016(E)Foreword iv1 Scope . 12 Terms and d

8、efinitions . 13 Purpose and motivation . 23.1 Purpose 23.2 Motivation . 24 Overview . 35 Annotation principles and requirements . 45.1 Principles inherited from the Linguistic Annotation Framework 45.2 Other general annotation principles 55.3 Principles specific to semantic annotation . 56 The metho

9、dological basis of SemAF . 76.1 Steps in the design of an annotation scheme 76.2 Metamodels . 86.3 Abstract syntax, concrete syntax and semantics 106.4 Steps forward and feedback in the design process .126.5 Optional elements in an annotation scheme 147 Overlaps between annotation schemes .157.1 Sem

10、antic and terminological consistency 157.2 Spatial and temporal relations as semantic roles 157.3 Events . 177.4 Discourse relations in dialogue . 188 Semantic phenomena that cut across annotation schemes 188.1 Ubiquitous semantic phenomena 188.2 Quantification . 188.3 Quantities and measures . 198.

11、4 Negation, modality, factuality, and attribution .208.5 Modification and qualification . 218.5.1 Modification and quantification 218.5.2 Qualification 228.5.3 Other issues .23Annex A (informative) An approach to the annotation of quantification in natural language .24Bibliography .28 ISO 2016 All r

12、ights reserved iiiContents PageBS ISO 24617-6:2016ISO 24617-6:2016(E)ForewordISO (the International Organization for Standardization) is a worldwide federation of national standards bodies (ISO member bodies). The work of preparing International Standards is normally carried out through ISO technica

13、l committees. Each member body interested in a subject for which a technical committee has been established has the right to be represented on that committee. International organizations, governmental and non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely wit

14、h the International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.The procedures used to develop this document and those intended for its further maintenance are described in the ISO/IEC Directives, Part 1. In particular the different approval criteria needed f

15、or the different types of ISO documents should be noted. This document was drafted in accordance with the editorial rules of the ISO/IEC Directives, Part 2 (see www.iso.org/directives).Attention is drawn to the possibility that some of the elements of this document may be the subject of patent right

16、s. ISO shall not be held responsible for identifying any or all such patent rights. Details of any patent rights identified during the development of the document will be in the Introduction and/or on the ISO list of patent declarations received (see www.iso.org/patents).Any trade name used in this

17、document is information given for the convenience of users and does not constitute an endorsement.For an explanation on the meaning of ISO specific terms and expressions related to conformity assessment, as well as information about ISOs adherence to the WTO principles in the Technical Barriers to T

18、rade (TBT) see the following URL: Foreword - Supplementary information.The committee responsible for this document is ISO/TC 37, Terminology and other language and content resources, Subcommittee SC 4, Language resource management.ISO 24617 consists of the following parts, under the general title La

19、nguage resource management Semantic annotation framework (SemAF): Part 1: Time and events (SemAF-Time, ISOTimeML) Part 2: Dialogue acts (SemAF-Dacts) Part 4: Semantic roles (SemAF-SR) Part 5: Discourse structures (SemAF-DS) Technical Specification Part 6: Principles of semantic annotation (SemAF Pri

20、nciples) Part 7: Spatial information (ISOspace)The following parts are in preparation: Part 8: Semantic relations in discourse (SemAF DR-core) Part 9: Reference (ISOref)iv ISO 2016 All rights reservedBS ISO 24617-6:2016INTERNATIONAL STANDARD ISO 24617-6:2016(E)Language resource management Semantic a

21、nnotation framework Part 6: Principles of semantic annotation (SemAF Principles)1 ScopeThis part of ISO 24617 specifies the approach to semantic annotation characterizing the ISO Semantic annotation framework (SemAF). It outlines the SemAF strategy for developing separate annotation schemes for cert

22、ain classes of semantic phenomena, aiming in the long term to combine these into a single, coherent scheme for semantic annotation with wide coverage. In particular, it sets out the notions of both an abstract and a concrete syntax for semantic annotations, mirroring the distinction between annotati

23、ons and representations that is made in the ISO Linguistic Annotation Framework. It describes the role of these notions in relation to the specification of a metamodel and a semantic interpretation of annotations, with a view to defining a well-founded annotation scheme.This part of ISO 24617 also p

24、rovides guidelines for dealing with two issues regarding the annotation schemes defined in SemAF-parts: a) conceptual and terminological inconsistencies that may arise due to overlaps between annotation schemes and b) the treatment of semantic phenomena that cut across SemAF-parts, such as negation,

25、 modality and quantification. Instances of both issues are identified, and in some cases, direction is given as to how they may be tackled.2 Terms and definitionsFor the purposes of this document, the following terms and definitions apply.NOTE In addition, the terms event and eventuality are used (a

26、s synonyms) as defined in ISO 24617-1 as something that can be said to obtain or hold true, to happen or to occur.2.1primary dataelectronic representation of text or communicative behaviourEXAMPLE Digital representations of text, transcriptions of speech, gestures or multimodal dialogue.Note 1 to en

27、try: ISO 24612 defines primary data as the electronic representation of language data. This definition is unsatisfactory for this part of ISO 24617 as semantic annotation may relate to non-verbal or multimodal data, such as stretches of spoken dialogue with accompanying gestures and facial expressio

28、ns, and even gestures and/or facial expressions without any accompanying speech.2.2annotationlinguistic information added to primary data (2.1), independent of its representationSOURCE: ISO 24612:2012, 2.32.3semantic annotationannotation (2.2) which contains information about the meaning of a segmen

29、t or region of primary data (2.1) ISO 2016 All rights reserved 1BS ISO 24617-6:2016ISO 24617-6:2016(E)2.4metamodelschematic representation of the concepts that are used in the analysis and description of the phenomena covered in annotations (2.2) and of the relationships between them3 Purpose and mo

30、tivation3.1 PurposeThe purpose of this part of ISO 24617 is to provide support for the establishment of a consistent and coherent set of international standards for semantic annotation within the Semantic Annotation Framework (SemAF). It aims to do so in three ways.First, by making explicit which ba

31、sic principles underlie the approach that has been followed in defining international standards in the SemAF parts that have been published so far (ISO 24617-1 and ISO 24617-2, ISO 24617-4 and ISO 24617-7), and in parts that are close to publication (ISO 24617-6) or in preparation (ISO 24617-8). Thi

32、s approach provides the Semantic Annotation Framework with methodological coherence and helps to ensure mutual consistency between existing, developing, and future SemAF parts.Second, by identifying overlaps between SemAF parts and indicating how such overlaps may be dealt with. Examples are the occ

33、urrence of temporal and spatial relations among semantic roles and of discourse relations between dialogue acts.Third, by identifying common issues that arise in various parts of SemAF (they are only partly covered in these parts, if they are covered at all) and, where possible, by giving directions

34、 as to how these issues may be tackled. Examples of such issues are polarity, modality, quantification, measures, qualification, veridicity, attribution and non-literal language use.3.2 MotivationSemantic annotation enhances primary data with information about their meaning. The state of the art in

35、computational semantics makes it unlikely that a single existing formalism for annotating semantic information would receive wide support from researchers and developers. Moreover, semantic annotation tasks often have the limited aim of annotating certain specific semantic phenomena, such as semanti

36、c roles, discourse relations or coreference relations, rather than annotating the full meaning of stretches of primary data. A strategy was therefore adopted in ISO TC 37/SC 4 to devise the SemAF standards in different parts, with separate annotation schemes for those classes of semantic phenomenon

37、for which the state of the art would justify the establishment of annotation standards; these schemes could be extended and combined over time, growing into a wide-coverage framework for semantic annotation.This crystal growth strategy has contributed significantly to the progress made in establishi

38、ng standardized annotation concepts and schemes supporting the development of interoperable resources, but it also entails certain risks:a) the annotation schemes defined in different SemAF parts are not necessarily mutually consistent, especially in the case of overlaps in scope;b) it may not be po

39、ssible to combine the schemes, defined in different parts, into a coherent single scheme with a wider coverage if they incorporate different views or employ different methodologies;c) some semantic phenomena do not belong to the scope of any SemAF parts but cannot be disregarded entirely in some par

40、ts, and this may result in these phenomena being unsatisfactorily treated.The methodological principles and guidelines provided in this part of ISO 24617 are designed to minimize these risks.2 ISO 2016 All rights reservedBS ISO 24617-6:2016ISO 24617-6:2016(E)With regard to the issue of mutual consis

41、tency between SemAF parts, it may be noted that ISO 24617-1 for annotating time and events and ISO 24617-2 for annotating dialogue acts are concerned with sufficiently distinct kinds of semantic information to allow their definitions to be established independently. Other SemAF parts, such as those

42、concerned with semantic roles, with relations in discourse and with spatial information show a certain amount of overlap in the information that they aim to capture, and the question therefore arises: can we ensure that the annotation schemes, defined in these parts, are mutually consistent?Mutual c

43、onsistency of SemAF parts relates to the possible integration of annotation schemes defined in different parts. For example, it would be desirable to use the ISO 24617-1 scheme (“ISO-TimeML”) for annotating time and events in combination with the ISO 24617-4 scheme for semantic roles, thereby annota

44、ting coherently not only the events identified in the data with their temporal properties, but also the way in which these events are related to their participants. Integrating these annotations with those of spatial information, using the ISO 24617-7 scheme for spatial information, would be another

45、 plausible and desirable step, given that time and space are intertwined with concepts relating to motion and velocity. More generally, the integration of SemAF parts would greatly enhance the significance of the individual parts; in the end, SemAFs crystal growth strategy of SemAF is only really us

46、eful if the annotation schemes defined in the various parts can grow into a single scheme with a wide coverage of semantic phenomena. Only then can it effectively support such applications as text-based question answering or extracting semantic information from text, and form the basis for automatic

47、ally recognizing semantic phenomena by means of machine-learning techniques. Clearly, this is only possible if the annotation schemes are mutually consistent (e.g. they use the same classification of event types), and are coherent whether, for example, temporal and spatial objects are viewed as even

48、t participants or as the circumstances of an event.With regard to the risk of unsatisfactory partial treatments of phenomena that are not among the core issues of any (current) SemAF part, it should be noted that some of these phenomena cut across several of these parts and are important for semanti

49、cs-driven applications. Negation, or more generally negative polarity, and quantification are two cases in point. Given that the aim in ISO-TimeML, for instance, is to support the annotation of events, of their relation to time, and of the temporal relations among temporal objects, it is desirable to be able to deal with sentences like the following:(1) John teaches every Monday.(2) Mary called twice this morning.(3) John rang home twice a day.Sentence (1) is about a set of “teach” events, each of which is related to a different