A semantic wiki to share and reuse knowledge into extended enterprise Inaya Lahoud
Davy Monticolo
Vincent Hilaire
Dpt. of Computer science University of Galatasaray Istanbul, Turkey
[email protected]
ERPI Laboratory University of Lorraine Nancy, France
[email protected]
Set laboratory University of technology of Belfort-Montbeliard Belfort, France
[email protected]
Abstract. This paper aims to present an approach in order to share knowledge between the business actors of the extended enterprise. This approach is based on a semantic wiki that access the organizational memory of our knowledge management system OCEAN. This principle allows us, firstly to share knowledge in order to facilitate the individual work, and secondly to reuse capitalized knowledge automatically. Keywords-Knowledge management, semantic wiki, multi-agents system
I.
INTRODUCTION
OCEAN [1] is a knowledge management system (KMS) whose main purpose is the management of heterogeneous knowledge during the design process of products. In this context, the analysis of major works on knowledge management highlights four main steps: the definition of knowledge, knowledge extraction, knowledge validation and knowledge reuse [2]. The first two steps allow assisting business expert to identify the needed knowledge for his project, to extract knowledge from business application, to annotate and save these knowledge in an organizational memory. These steps aim to enable the diffusion and reuse of knowledge in order to make them reachable to actors. This diffusion represents an exchange or transfer of knowledge between a source and a destination. This exchange can be done indirectly by using information and communication technology. In addition, diffusion, which naturally occurs in limited circles of acquaintances in a conventional company, has become even more difficult in the context of an extended enterprise. A KMS must therefore allow access and filtering of knowledge in order to make this diffusion effective. Our knowledge diffusion approach is based on the design, implementation and use of a semantic wiki that allows validation, sharing, reuse and evaluation of knowledge stored in organizational memory. The advantages of a semantic wiki reside in its semantic structure. This structure, which is readable and handled by a computer, improves the possibilities for navigation, query, sharing and reuse of knowledge. The rest of the paper is structured as follows: Section 2 presents some existing semantic wiki and their importance in knowledge sharing process. Then we present in the section 3 our overall approach of
our wiki based on multi-agents system. We explain in this section the features of our wiki. Section 4 concludes this article. II.
SEMANTIC WIKI FOR SHARING KNOWLEDGE
We discussed in the introduction that knowledge are distributed across the entire network of the extended enterprise. This one faced by problems of management of heterogeneous and distributed knowledge. Thus, the question is to know how to share knowledge between the different enterprises and between business actors of the same enterprise. Semantic wiki seems to be a good solution to this kind of problem. It has proven its ability to manipulate and share knowledge simply and rapidly. A wiki is a web site that allows collaborative distant creation of information and editing of hypertext content. Leuf & Cunningham[3] were the first to propose a web site where people could create, modify, transform and link pages within their browser and in a very simple way. Wikis have become popular tools for collaboration on the web [4], and many active online communities employ wikis to publish or exchange information. For most wikis, public or private, the primary goals are to organize the collected information and to share it. Wikis are usually viewed as tools to manage online content in a quick and easy way, by editing some simple syntax known as wikitext [5]. Schaffert [6] enumerates the specifications of a wiki system: It allows the editing via a browser; It has a simplified wiki syntax i.e. simplified hypertext format usable by all the internet users; It manages a rollback mechanism i.e. it is able to versioned the changes in the content each time they are stored; Its access is unrestricted, everybody can write in the wiki; It manages the collaborative editing i.e. if someone create a article, everybody can extend this article; It proposes a strong linking, all the pages of the wiki are linked with each other using hyperlinks; It has a search function over the content of
all pages stored; It allows the uploading of different content like documents, images or videos. Taking into consideration all these properties, Wikis seems as a good candidates for collaborative knowledge engineering based on Web2.0 social networks [7]. Indeed new research works [6], [8] propose wikis to exchange knowledge. Knowledge can be seen as information with an added context and value that make it usable within that very context. Knowledge can also be seen as what places someone in the position to perform a particular task by selecting, interpreting and evaluating information depending on the context [9], [10]. However a serious obstacle for the development of Semantic Web applications is the lack of formal ontologies and knowledge. Indeed, one of the main reasons of this is the rather high technical barrier for using Semantic Web [11] technologies that deters many domain experts from formalizing their knowledge. On the other hand, wiki systems are becoming more and more popular as tools for content and information management. Much information is nowadays available in systems like Wikipedia [12], [13]. Unfortunately, this vast information is not accessible for machines. If a small amount of this information could be formalized to become knowledge, then, wiki systems could provide improved interfaces, advanced searching, and navigation facilities. Nevertheless, several analyses[14], [8] of traditional wikis as shown that they are not enough structured, and it’s difficult to navigate and to find the relevant information. Besides, the wiki markup language (WikiML) used by most wiki engines makes internet users reluctant to contribute to the wiki. One solution to perform the ideas creation, evaluation and navigation inside wikis is to use technologies from the Semantic Web [15] to formalized information, content, structures and links in the wiki pages. These Wikis would take consideration of the semantic in their content management and become Semantic Wikis[16], [17]. “Semantic Wiki” systems aim to combine “traditional” wiki systems with Semantic Technology. A semantic wiki is a system that allows collaborative authoring, editing and linking of pages, but also the authoring and adding semantics to the data on the wiki itself. The significance of semantic wiki is that it may contain machine-readable content and structure, which will improve the possibilities to browse, query, share and reuse the knowledge. Völkel [10] list a number of requirements for semantic wikis: usability, expressiveness, flexibility, scalability, interchange and compatibility.
Since 2004, with the development of Platypus Wiki [18], many sematic wikis have been created as: SweetWiki [14], MaknaWiki [19], IkeWiki [6], OntoWiki [20], Shawn [21], Rise [22], Semantic MediaWiki [12], WikSar [15]). These wikis provide the possibility to store in RDF format and edit its content. These wikis was based on « wikitology » model [22]. That means wiki pages are considered as concepts and links as relations or attributes. Wiki can be considered in this case as an ontology with concepts, relations and attributes, in other words, it compose the ontology. Our approach is based on another model, inspired from WikiDesign [23]. This wiki, as SweetWiki and IkeWiki, uses ontology to edit the content of pages. Nevertheless, WikiDesign has been designed to: (i) be associated with a knowledge management system and (ii) facilitate the creation and validation of new knowledge. WikiDesign uses domain ontology (OntoDesign) to edit its pages. Thus, wikis can be considered as a good candidate for collaborative knowledge management based on Web2.0 technologies [7]. We present in the next section the architecture of our semantic wiki. III.
WIKI ARCHITECTURE
A. Overview The semantic wiki is based on the contents of organizational memory. This organizational memory is built by extracting knowledge in our OCEAN system. This extraction is being handled by a multiagents system [24]. Contextualize the data retrieved by agents transforms them into information. This information is stored in RDF files. Thus, RDF files constitutes the organizational memory. We must note that RDF files contain only information. This information is transformed into knowledge when it is interpreted by a human[25]. In our case, the information stored in RDF files becomes knowledge after a validation by a business actor in the semantic wiki. Figure 1 shows the three levels of our semantic wiki that are: the web layer, knowledge processing layer and knowledge persistence layer. These levels are managed by a multi-agents system (MAS). Our semantic wiki allows to disseminate the knowledge of organizational memory. To access this knowledge, business actor connects to wiki and launches a search (query) as needed. The agents, that are responsible to search in the organizational memory the corresponding RDF files, process this request. The result of the research will be presented in the wiki interface as shown in Figure 1.
Web layer (user interface)
M A S
Knowledge processing layer (SPARQL Query Process)
Knowledge persistence layer (organizational memory (RDF files))
PREFIX rdf: PREFIX rdfs: PREFIX owl: PREFIX xsd: SELECT ?x ?nameRdf ?z WHERE { ?x "speed" .}
S 525 545
Figure 1: Global architecture of our semantic wiki
We consider in this paper that a wiki page corresponds to an RDF file in the organizational memory. As we work in an industrial field, each page contains a draft describing the design of a mechanical product as the design of an engine for a specific car or designing an office chair. In the next section, we detail the various features of this wiki. For each feature we will show the mechanism of its functioning and the agent that assure it.
Our wiki offers a semantic search engine to query and reason on the organizational memory. Semantic search is used to improve search accuracy by understanding the purpose of research and the textual meaning of business actor query. The purpose of this research is to have results that are more relevant for business actor. For example, if you ask "How fast runs a jaguar", the wiki system must understand that you do not talking about the car brand, and will analyze the terms of the question to give you the best answer.
B. Wiki features The main objective of our semantic wiki is to disseminate knowledge within the extended enterprise by allowing business actors navigation and semantic search within the wiki pages. This wiki provides then the following features: semantic search, classification, representation, modification and evaluation of knowledge (Figure 2).
Our idea for semantic search was to make crosses of information between different wiki pages and accumulate them with the personal information of business actors. This crossing is done by linking items (inheritance for instance). Personal information is defined in the user profile page. Linking with personal information allows giving results close to the centers of interest of actor which limits the number of irrelevant one.
1) Semantic search
Semantic wiki Representation Classification Classification system
Evaluation Search results presentation
User profile Identification
Research query
Created projects
interface
actor
Role
Consulted projects
Wiki features page
rate
Domain of interets
Searched words
Legend:
actor
Feature
Ranking User Profile
Modification
agents actions
page modify values
Interactions between features
actor
Organizational memory
page Organizer Agent
Role Agent
Validation Organization
page
Expert Agent
Wiki pages
Announcer Agent
Figure 2: wiki features
To perform this semantic search, expert agent monitor semantic wiki interface and reacts to each entry of terms in the interface (Figure 3). This reaction involves contacting the announcer agent. The announcer agent identifies the terms of business actor query and generates the corresponding SPARQL queries. To enrich these queries, this agent
asks for the user profile information from the database. This allows to have more specific and relevant results. The result of semantic search is a list of wiki pages. However, this list is returned without any ranking, so to present them in a structured way we apply a classification approach. This function is the subject of the second feature of our wiki.
Figure 3: Diagram of sequence for semantic search feature
2) Classification To make the presentation of query results more relevant to business actor we present a classification approach that display them by the most important to him (Figure 4). In this approach, we use two classification methods: the user's profile and evaluation of wiki pages. The first is to sort the results by applying different types of filters such as 1) the role of the user, 2) areas of interest defined in his profile, 3) projects that he consults, 4) projects that he creates, and 5) the history of his research in the wiki. The second method is to sort the results by the maturity index. This index is the number of positive evaluations of the wiki page. We obtain this index by dividing the total score (rate) by the number of voters (nbRaters). More the maturity index is high more it appears in the beginning of the results list. This maturity index
is represented by yellow stars as shown in Figure 4. We explain the approach to evaluate a page in the evaluation feature section. To realize this feature, the organizer agent receives the results not classified corresponding to the query of business actor. This agent asks the profile of this actor to know his areas of interests, role, etc. When the organizer agent receives the profile of the actor, he launched a second request to know the maturity index of each wiki page from the organizational memory. Once the organizer agent has the information he need, he classifies the pages and returns a list of wiki pages classified. The organizer agent will send this list of classified items to announcer agent who present it to the business actor.
Figure 4: Classification approach
Figure 4 shows an example of the classification scheme in which the user “Davy” occupies the role of “designer”. This user is interested in the areas of “industry, mechanical, maintenance, aerospace, nuclear, hydraulic, vehicle”. When he looks in the system for the word “speed”, for example the classification system sorts the results by the mechanical field first, because this word also appears in the areas of food, security, sports, etc. 3) Representation The presentation feature consists of two tasks performed by the announcer agent. This agent ensures, on the one hand, the presentation of classified results of semantic search to business actor. On the other hand, announcer agent allows the actor to manage its profile. In this profile, he can define the different roles he has in the project and areas of interest. According to this information, our wiki filter the query results and present only those
linked to his profile. 4) Evaluation & Modification We explained in the previous section that the agent presents the search results in the form of classified wiki pages. However, business actor can navigate the wiki pages. When he clicks on a page “role” agent directs him to another page displaying the knowledge of this page. Our wiki allows the actor two choices: either evaluate the page or modify it. The modification in our wiki corresponds to change the values of the knowledge present in the page by the business actor. In our Wiki, knowledge is also subject to an evaluation process (Figure 5). Business actors make this assessment. In this case, the actor gives a score between one and five for each knowledge. This score means that knowledge is seen as relevant if it is close to five and the opposite if it is close to one.
Figure 5: interface d'évaluation des connaissances
The knowledge evaluation process, as in the case of modification, is triggered when the expert agent detects an evaluation action made by the business actor. In this case, expert agent takes the rating set by business actor and sends it to the “role” agent. This agent research concerned wiki page from the organizational memory and load it. Agent made a number of changes; first, he updates the overall rating of this wiki page by adding the note putted by business actor. Then he increments the number of voters on this page and finally he calculates the index of maturity of this page by dividing the total score on the number of voting. When the role agent finishes the modifications, he communicates with the announcer agent to resubmit the concerned wiki page with his new maturity index.
REFERENCES
[1]
[2]
[3] [4]
In the evaluation process, we consider that business actor approves a page when he assigns a positive feedback. Moreover, when he refuses, the page gets negative feedback. Thus, knowledge that has just been created has 100% positive feedback. As the assessments assigned by business actors, the percentage (maturity index) may decrease or increase. If the index is less than 25% of positive evaluation, knowledge will be considered as nonuseful and will be proposed to an expert for deletion. IV.
[5]
[6]
CONCLUSION
In this paper, we presented the module "diffusion and knowledge reuse" of our OCEAN system. We used a semantic wiki based on MAS to disseminate knowledge. The architecture of the wiki and the features it provides are explained in this paper. This wiki is used to evaluate, modify, and share knowledge of our organizational memory. Our Wiki is particularly useful for user groups and communities working together to create new knowledge for mechanical projects and maintain this knowledge. It also offers a user-friendly way to build and maintain common vocabularies to enterprise’s employees.
[7]
[8]
I. Lahoud, D. Monticolo, et S. Gomes, « OCEAN: A Semantic Web Service to Extract Knowledge in E-Groupwares », in 2010 Sixth International Conference on Signal-Image Technology and Internet-Based Systems (SITIS), 2010, p. 354 ‑362. I. Lahoud, D. Monticolo, V. Hilaire, S. Gomes, et E. Bonjour, « A Multi-Sources Knowledge Management System », in Information Control Problems in Manufacturing, 2012, vol. 14, p. 1177‑1183. B. Leuf et W. Cunningham, « The Wiki way: quick collaboration on the Web », 2001. A. Majchrzak, C. Wagner, et D. Yates, « Corporate wiki users: results of a survey », in Proceedings of the 2006 international symposium on Wikis, 2006, p. 99–104. A. V. Singh, A. Wombacher, et K. Aberer, « Personalized information access in a wiki using structured tagging », in On the Move to Meaningful Internet Systems 2007: OTM 2007 Workshops, 2007, p. 427–436. S. Schaffert, « IkeWiki: A Semantic Wiki for Collaborative Knowledge Management », in 15th IEEE International Workshops on Enabling Technologies: Infrastructure for Collaborative Enterprises, 2006. WETICE ’06, 2006, p. 388‑396. D. Richards, « A social software/Web 2.0 approach to collaborative knowledge engineering », Inf Sci, vol. 179, no 15, p. 2515–2523, juill. 2009. D. Vrandecic et M. Krötzsch, « Reusing ontological background knowledge in semantic wikis », in Proceedings of the 1st Workshop on Semantic Wikis, From Wiki To Semantics, AIFB, ESWC2006, Budva, Montenegro, 2006.
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
T. W. Malone, K. Crowston, et G. A. Herman, Organizing business knowledge: the MIT process handbook. MIT press, 2003. M. Völkel, M. Krötzsch, D. Vrandecic, H. Haller, et R. Studer, « Semantic wikipedia », in Proceedings of the 15th international conference on World Wide Web, 2006, p. 585–594. T. Berners-Lee, J. Hendler, et O. Lassila, « The Semantic Web », Scientific American, p. 29‑37, mai-2001. M. Krötzsch, D. Vrande\vcić, M. Völkel, H. Haller, et R. Studer, « Semantic wikipedia », Web Semant. Sci. Serv. Agents World Wide Web, vol. 5, no 4, p. 251–261, 2007. M. Krötzsch, D. V. Denny Vrandecic, et M. Völkel, « Wikipedia and the semantic webthe missing links », in Proceedings of Wikimania 2005, 2005. M. Buffa, F. Gandon, G. Ereteo, P. Sander, et C. Faron, « SweetWiki: A semantic wiki », Web Semant. Sci. Serv. Agents World Wide Web, vol. 6, no 1, p. 84–97, 2008. D. Aumüller et S. Auer, « Towards a semantic wiki experience–desktop integration and interactivity in WikSAR », in Semantic Desktop Workshop, Galway, Ireland, 2005, p. 2005–2012. J. Fischer, Z. Gantner, S. Rendle, M. Stritt, et L. Schmidt-Thieme, Ideas and improvements for semantic wikis. Springer, 2006. A. Souzis, « Building a semantic wiki », Intell. Syst. IEEE, vol. 20, no 5, p. 87–91, 2005. S. E. Campanini, P. Castagna, et R. Tazzoli, « Platypus wiki: a semantic wiki wiki web », Semantic Web Appl. Perspect. Proc. 1st Ital. Semantic Web Workshop, 2004. K. Dello, E. P. B. Simperl, et R. Tolksdorf, « Creating and using semantic web information with makna », in First Workshop on Semantic Wikis, From Wiki to Semantics, Budva, Montenegro, 2006, p. 1. S. Auer, S. Dietzold, et T. Riechert, « OntoWiki–A tool for social, semantic collaboration », in The Semantic Web-ISWC 2006, Springer, 2006, p. 736–749. D. Aumüller, « SHAWN: Structure helps a wiki navigate », in Proceedings of the BTWWorkshop WebDB Meets IR, Karlsruhe, Germany, 2005. B. Decker, E. Ras, J. Rech, B. Klein, et C. Hoecht, « Self-organized reuse of software engineering knowledge supported by semantic wikis », in Proceedings of the Workshop on Semantic Web Enabled Software Engineering (SWESE), Galway, Ireland, 2005. D. Monticolo et S. Gomes, « WikiDesign:Collaborative Knowledge
Evaluation with a Semantic Wiki », Int. J. ECollab., vol. 7, no 3, p. 31‑42, 33 2011. [24] I. Lahoud, D. Monticolo, V. Hilaire, et S. Gomes, « A Metamodeling and Transformation Approach for Knowledge Extraction », in Networked Digital Technologies, R. Benlamri, Éd. Springer Berlin Heidelberg, 2012, p. 54‑68. [25] F. Gandon, « Distributed Artificial Intelligence and Knowledge Management: Ontologies and Multi-agent Systems for a Corporate Semantic Web », thesis, 2002.
