胜任力模型在定义方面界定胜任特征外文翻译（可编辑）

胜任力模型在定义方面界定胜任特征外文翻译（可编辑） 胜任力模型在定义方面界定胜任特征外文翻译 外文翻译 Competency model in a semantic context: Meaningful competencies Material Source:Vrije Universiteit Brussel Author:Christiaens In this paper, we will propose our ideas for a semantically ready competency model. The model will allow semantic enrichment on different levels, creating truly meaningful competencies. The aim of this model is to provide a flexible approach for reuse, matching, interpretation, exchange and storage for competencies. Our competency model is based on the DOGMA ontology framework and the proposed IEEE standards RCD and SCRM. We will focus on the model itself and how semantics can be applied to it as these elements form the basis for any kind of processing on them. Keywords:competence,competency, RCD, semantics, DOGMA, e-HRM, HRM, occupation, ontology, Semantic Web. On the highly volatile job market that we are living in at this moment, getting a good match between a CV and a job opening is a key problem. The main element in this matching process is shifting from monolythic function titles to functions built with minimalistic, descriptive and highly-reusable competency building blocks. Such a bunch of small information pieces put together allows for a far more detailed and closely fitting match. Both a person and a function can be seen as a grouping of competencies and fitting one with the other becomes a matter of comparing these collections of building blocks. What is still missing in this picture, is the exact view of a competency and a group of competencies. What identifies a competency? How do you describe it? How can we identify the intended meaning of a competency? And how do you put competencies together in a meaningful and clear manner? These are the kind of questions that we try to answer in our model of meaningful competencies. A short look at the current state-of-the-art with these questions in mind results only in partial answers. The HR-XML consortium provides a schema for competency description, however the main focus seems to be on the possibility to capture evidence and provide weighting. They provided an element to store semantics, but only limited to taxonomy. Claude Ostyn created the reusable competency definition RCD which is expected to be accepted as an IEEE standard soon. This standard provides the minimal elements id, title, description and a structured free-text definition to describe a competency. A metadata field allows the introduction of semantics, however there is no further specification about how exactly semantics should be introduced and what they look like. Ostyn also proposed the Simple Reusable Competency Map SRCM standard. This pending IEEE standard provides a way to capture possible parent-child/sibling relations between RCDs. This way different communities of practice can use the same RCDs, but group them together differently. The limitation of this approach is in the fact that the relation types are limited to parent-child or sibling. Our work is based on the current situation, but forms a model which allows rich semantics on different levels. In this introduction we described the theory which served as the basis for our competency model.We give a brief overview of the DOGMA framework for ontology engineering in Section. Our model was inspired by this framework, and we also make use of it for formalizing semantics. In Section we present our model and how meaning fits in this picture. Finally, in section, we summarize our further intentions. 2 The DOGMA Framework DOGMA is a research intitiative of VUB STARLab where various theories, methods,and tools for ontologies are studied and developed. A DOGMA inspired ontology is based on the classical model-theoretic perspective and decomposes an ontology into a lexon base and a layer of ontological commitments. This is called the principle of double articulation. A lexon base holds multiple intuitive conceptualisations of a particular domain.Each conceptualisation is simplified to a"representation-less"set of context-specific binary fact types called lexons. A lexon represents a plausible binary fact-type and is formally described as a 5-tupleV, term1, role, co-role, term2, where V is an abstract context identifier, lexically described by a string in some natural language, and is used to group lexons that are logically related to each other in the conceptualisation of the domain. Intuitively, a lexon may be read as: within the context V, term1 also denoted as the header term may have a relation with term2 also denoted as the tail term in which it plays a role, and conversely, in which term2 plays a corresponding co-role.Each context, term-pair then lexically identifies a unique concept. A lexon base can hence be described as a set of plausible elementary fact types that are considered as being true. The terms in the lexon combined with the context resulting in 2 concepts.By also conceptualizing the role/co-role, the lexon becomes a language- and context independent metalexon. Any specific application-dependent interpretation is moved to a separate layer, //0>. commitment layer. The commitment layer mediates between the lexon base and its applications. Each such ontological commitment defines a partial semantic account of an intended conceptualisation. It consists of a finite set of axioms that specify which lexons of the lexon base are interpreted and how they are visible in the committing application, and domain rules that semantically constrain this interpretation.Experience shows that it is much harder to reach an agreement on domain rules than one on conceptualisation. E.g., the rule stating that each car has exactly one license plate number may hold in the Universe of Discourse UoD of some application, but may be too strong in the UoD of another application. A full formalisation of DOGMA can be found in De Leenheer,Meersman, and de Moor 3 Meaningful Competencies 3.1 Competency Model As we explained in the previous sections, the DOGMA architecture divides the complexity over three levels: lexonbase, commitment layer and application layer. We were inspired by this approach to construct our competency model. The total overview on the model can be seen in Figure. The lowest level is the RCD or competency repository, the middle level is the reusable competence layer and on top are the users of the competency data, the applications.We adopt the definition of competency as given by the HRXML consortium: A specific, identifiable, definable, and measurable knowledge, skill, ability and/or other deployment-related characteristic e.g. attitude, behavior, physical ability which a human resource may possess and which is necessary for, or material to, the performance of an activity within a specific business contextA competency is used as the smallest unit of capability in our model a highly ?exible, modular building block. A competence is a structured set of competencies. Note that in our model a competency can only be used to construct competences. We will give a more detailed description of every layer in the following subsections. Competency Repository. The lowest level is a repository filled with RCDs. These RCDs can be obtained by automatic mining e.g., from O*NET, by human input or both. Any RCD present in the repository can be linked with an arbitrary relation to any other RCD. These relations have no interpreted meaning in the competency repository layer other than the one in the mind of the human creator of the relation.We call them plausible relations, as they represent a fact that might or might not be true. As we do not restrict ourselves to the "is composed of" relation, we can build a more semantically rich set of competencies. For instance, an RCD"can drive" can be related with an RCD "can read road signs"using an "assumes"-relation instead of a "requires"-relation to indicate a softer constraint between the two RCDs. The RCD is built using a dedicated web interface still to be created which will encourage the reusability of the competencies. For instance, if the repository already contains an RCD "can drive a car", and the user creates a new RCD" can drive a truck", the system will inform the user of the existing RCD and encourage him to group these related RCDs with a relevant relation. Reusability is a strong point to interoperability, as the use of the exact same objects competencies implies a shared understanding. Furthermore, we will allow the user to semantically tag or DOGtag the RCD with a part of the domain ontology. The system will propose concepts and relations between them from the ontology as semantic tags and the user can select the exact one or multiple ones. In this way, an RCD can be annotated with semantics, ranging from one simple concept to complete semantic graphs. We will make use of the already entered fields of the RCD title, description and definition to apply some heuristic search for possibly related concepts in the ontology. This metadata can be stored in the metadata field of the RCD. We will make use of the most commonly known formats for such description, like RDFS and OWLNext to this information we will also store standard Dublin Core metadata. Competence Repository. The competence repository creates a committed view on the RCD repository. Applications can make a selection of the RCDs and the plausible relations between them and commit to them. This means that the application fixes this structured set of RCDs to be truthful for its own application context. At this point the application commits to using exchanging,matching,. that specific competence built from those specific RCDs and their relations. As such it also locks the interpretation of the plausible relations, and thus the relations are no longer simply plausible, but actually factual. The application can now provide an explicit formalization of the relation meaning e.g., if it has to match with other applications or simply leave it as a term with implicit meaning e.g., if it only uses the competence internally. The user can describe the competence with a title. For instance, she creates the competence"can drive"containing the RCDs"can drive a car"-requires-"can understand traffic signs". The contained RCDs and their inter-relations form the meaning for the competence. In another meaning other RCDs the competence ”can drive” could have another interpretation stored in the contained RCDs of that competence. Again, we will build a dedicated web interface for constructing these competences. It will encourage the user in several ways see the previous subsection to reuse existing competences. At this layer we do not introduce additional tagging functionality. Searching for existing competences can be done by using the semantics inside DOGtag and outside RCD relations that are part of the competence. As the picture shows, a competence is more than data from the competency repository. Extra fields are also added to each RCD to further specify the competency. Examples of such fields are proficiency level, level of importance, level of interest,. It is important to note that the values of these fields are not yet fixed, as this would hinder reuse. Application Layer. The application layer holds all applications that will use our competency model and its data. Each application uses the same repository competence and competency. The applications work with application profiles, collections of competences from the competence repository. The users of these applications are responsible for filling both repositories with relevant data. However, one type of application can be completely dedicated to adding data, while another only uses existing data. The application collects required competencies and populates the extra field values. It is also responsible for setting the expertise level of the competence. The expertise level specifies at what level the competence needs to be.With level we do not simply mean a number e.g., on some further unspecified scale or a symbol e.g., good, bad, medium, but a real definition of the expertise level. Using semantic annotation DOGtag the expertise level can be linked to the domain ontology, and thus be expressed formally.We constrain this expertise level by stating that it can only make the competence more specific. For instance "can drive a car"can have an expertise level"can drive a car in the dark", but not"can drive a truck" as the first level is a specification of the competence,while the second is not if the used ontology does not state that"truck"-is a-"car". 译文 胜任力模型在定义方面:界定胜任特征 资料来源: 布鲁塞尔自由大学 作者:克里斯蒂安 在本文中,我们对胜任力模型语义进行研究,旨在研究胜任特征不同层次的语义,创造出规范的胜任特征。本研究的目的在与提供一个可重复使用,能够实际运用的胜任力模型。我们的胜任力模型是基于胜任力辞典和SCRM,专注于模型本身,以及如何在实际工作中运用,因为这些因素构为任何一种胜任力模型奠定了基础。 关键词:胜任力,胜任特征,RCD,语义学,DOGMA, e-HRM, HRM 1绪论 我们生活在就业市场大幅波动的时代,要在员工和岗位之间获得很好地匹配是一个关键性的问题。该匹配过程的主要从具有功能标题,以兴建简约的、描述性的和可重复使用的胜任力素质模型,如同信息技术里,将一个小的信息碎片整理成一束更为详细密切拟合的匹配。一个人的胜任力素质可以视为多种能力组合在一起的系列。 我们缺少一个胜任力和胜任特征群的通用。如何辨别一个能力?如何描述它?我们如何可以确定一项胜任力真实的意义?如何用正确科学的方式提出胜任力模型?这正是我们在界定的胜任力素质模型所要回答的问题。 目前来看,当前最先进的研究仅有部分回答了这些问题。HR-XML联盟提供了一个胜任力的定义,然而其重点在于如何构建胜任力。他们提供了一个胜任力素质辞典,但类型不多,只有通用素质。Claude Ostyn创造了一种更使用,能够被人们接受的可重复使用的胜任力辞典。该辞典规定了胜任力素质的基本要求,描述描述的一项胜任力需包含标题和描述性的语句。其关于胜任力的介绍,并没有进一步的说明应当如何正确地使用。同时,Ostyn还提出了简单可重复使用的 胜任力素质模型图。这解决了胜任力与胜任力模型之间的关系。此方式可以在不同情况下使用,又可以以不同的方式组合在一起。这种方法的局限性在于只能够用在特定的职位上。由于日常生活中职位是丰富多样的,需要形成一种能够满足不同需求的胜任力素质模型。 以上的研究为接下来描述胜任力模型奠定了理论基础。该胜任力模型来自于上述的胜任力辞典,同时也对它进行更为规范的定义。在第二节,我们提出胜任力素质模型在模型中是如何定义的。最后一部分,我们进一步进行总结。 2 胜任力素质 “素质模型”原本的含义是布鲁塞尔自由大学立体星象馆对各种理论、方法与本体加以研究和发展地方的工具。胜任力辞典基于古典模型理论视角,将模型分解成词语并且分层。这就是所谓的双重衔接的原则。 任何特定的胜任力素质都需要进行解释。名词和名词解释之间的联系都需要科学的研究及验证。每次这样的定义由于研究的局限性而导致其不够完整。胜任力辞典包括一组的特定的词汇对胜任力素质进行解释,以及它们是如何表现的。实践证明,这更加难以是胜任力素质规范化。例如,如果实际生活中,每辆汽车都拥有一个车型号,就可以很快的知道那是一辆什么样的车。因此,一个完整的胜任力素质辞典被De Leenheer, Meersman和de Moor提出。 3胜任力意义 3.1胜任力模型 正如我们在上一节所解释,一个完整的胜任力素质由三个部分组成:名 称、定义和级别。我们根据这种方法来构造出胜任力素质模型。在图中,可以看到模型的整体。最底层是技能知识库,中间层是可重复使用的能力层,顶层是胜任力数据库及运用。我们采用由HRXML联盟定义的胜任力:特定的、可以辨认的、 可定义的和可衡量的知识、技能、能力或其他相关的特性(例如,态度、行为、体能 以及人力资源所能控制的具体事务中的绩效。胜任力模型由多个胜任力素质组成。一个胜任力素质是胜任力模型的一部分。值得注意的是,在文中所述的胜任力模型仅可用于构建胜任力。我们将在以下小节中做更详细的说明。 名称库。最底层是技能知识库。这些知识技能可以从O*NET自动获得、人工输入或两者兼而有之。可以在库中连接到任意相关的任何其他技能知识条,用来解释技能知识库中的词组。因为他们代表了一个可能或可能不是真正的事实,我们要求他们看似合理的关系。正如我们不会限制自己到特定的关系中,可以构建一套词义更丰富的胜任力。例如,一个技能知识库中的词组可能会与另外两个词组之间产生联系。 技能知识库是使用专门网站界面,鼓励用户使用统一的可重复使用的词组构成的胜任力素质模型。例如,如果库中已经包含了一个“可以开车”,用户可以创造出一个新的词组“能开卡车”,系统会提示已存在的帮助提示,并鼓励他们运用这些词组建立相关联系。可重复性是一种词汇的互用性,指在对完全相同的对象能力进行表述时所使用的同样一种词汇。此外,用户可以使用语义标签,该系统将提出的概念及它们之间的关系进行研究,并从库中为用户提供可以选择准确的一个或多个词组。这样,一个词组可以标注定义,用一组简单的词汇来完成素质模型图。我们将已通过的词汇都存在系统的资料库中,为用户搜索资料库提供了可能。系统将使用频率最高的词组下来。之后在下一个用户需要构建胜任力 模型时提示胜任力素质的词组。 定义库。创造胜任力词库的基础是定义库。系统可以根据用户的使用信息重新定义胜任力素质,意味着程序能够自动修改定义,这方便了词组名称与胜任力素质定义之间互相交换和更换。