Remoter: an Operational Knowledge Management System for telecommunication operators

Transcription

1 Remoter: an Operational Knowledge Management System for telecommunication operators Giulio Valente 1 and Alessandro Rigallo 2 Abstract. This paper describes an Operational Knowledge Management System (OKMS) called Remoter. This OKMS has been used in a telecommunication context, to support the Telecom Italia s technicians during their day-to-day activities. Remoter integrates multiple technologies to support telecommunication operators who need operational-knowledge to perform their activities. The operational knowledge (OK) could be defined as a set of procedures or norms used by telecommunication operators to construct a genuine strategy tailored to the peculiarities of a given situation. Due to the approach followed, Remoter could be used in several different Telecommunication contexts. In particular, it has been used and tested in the Asymmetric Digital Subscriber Line (ADSL) context, where operators have to rapidly react and reach the best decision. The aim of Remoter is to enable ADSL technicians to share, capture and apply their collective OK to take optimal decision in real time. 1 Introdution The importance of competence and knowledge in Telecommunication context is constantly growing today, due to several issues (complexity of the hardware and software systems, customer retention, time-to-market etc.). Nowadays, companies like Telecom Italia are increasingly distributed world wide. As a result, it is more difficult to collaborate with ones peer. Therefore, they have recognized that in order to remain competitive, they must efficiently and effectively create, locate, capture and share their OK in a innovative way. In particular, the offer and maintenance of new services is the Telecommunication field where the ability to reuse and manage competence and knowledge is a key factor to the success in a competition. Let us consider the ADSL service. In this context, two groups of technicians have been identified: 1. help desk/back office technicians: people working in an office, whose main activity is to help the on-the-field technicians where some trouble related to the ADSL context occurred; 2. on-the-field/front-end technicians: technicians mainly involved in provisioning and maintenance activities, for example ADSL components installation, troubleshooting, etc. When a customer wants to install a new ADSL component or some troubles arise, call the ADSL customer care center of Telecom Italia who arranges an appointment between the customer and a front-end technician. Then the front-end technician goes to the customer in order to perform his work request. In some cases, on-the-field technicians call help desk/back office technicians for advice. However, 1 Dipartimento di Informatica Universita di Torino, Torino, Italy valenteg@di.unito.it 2 Telecom Italia, Torino, Italy alessandro.rigallo@telecomitalia.it both the back office and on-the-field technicians usually access to technical documents, knowledge spread on different web-sites in order to perform their activities. They often do not have on their desk the last version of a technical manual about an ADSL component and waste their time searching it on the web. At time, technicians in two different regional area have two different way to carry out an activities or to solve a trouble. Therefore, they need to share, fast access and manage knowledge regarding the ADSL services, especially when they are new to the job. We have developed an OKMS, called Remoter, in order to help technicians during their day-to-day activities of provisioning and assurance of the ADSL service. Remoter allows ADSL operators to manage, share and apply their collective OK to take optimal decision in real time. A brief description of the Remoter Architecture will be given in the following section. Section 3 discusses our motivation developing Remoter system and section 4 describes Remoter s user roles and functionalities. In section 5, we show the OKM phases in Remoter, and finally sections 6 and 7 discuss related and future work on improving the Remoter system. 2 Remoter Architecture The proposed system is a three tier web-based architecture, which allows managing and sharing the OK (see fig. 1). The OK is mainly based on individuals competence and experience, namely some sets of procedures used by people to construct a genuine strategy tailored to the peculiarities of a given situation (norms, technical documents, procedures, personal experience, links to web-site etc...). The presentation layer contains some GUI modules that provide the userinterface. This layer provides not only a graphical interface so that users can interact with the application, insert data, and view the results of requests, but it also manages the manipulation and formatting of data once the client receives it. A client device can communicate with Remoter using a web-browser with Java T M technology. In fact, the presentation layer consists of either a set of HTML with JavaScript or ASP and JSP pages. The second tier is a Web application server running the logic of the disciplined development process on the Web server. In particular, it supports some Logical Modules, which have been developed integrating different technologies as Autonomy T M Search and Retrieval Engine [7], egain T M Knowledge [6], Snitz Forum [8]. Autonomy Search and Retrieval Engine allows content to be searched and presented with summaries and hyperlinks to similar information, automatically and in real-time. egain T M Knowledge is a conversational Case-based Reasoning (CCBR) shell. Conversational Case-based Reasoning is a problem-solving paradigm that is able to utilize the specific knowledge of previously experienced, similar problem situation (case) to solve a new problem [10]. egain T M Knowledge stores knowledge

2 in the form of cases in a case base. Snitz Forum is a freeware interactive discussion environment that allows different people on the Internet or an Intranet to leave messages for each other, and then to reply. Moreover, the middle-tier is responsible of the users authentication and causing client devices to display the applications. A database server and a document repository constitute the third tier. The database server is used to store either messages posted by the users or the egain T M case base. The document repository is used to store all knowledge records (i.e. technical documents, pieces of technical information, norms, software etc...). The benefits of a three-tier architecture are significant. Being a web service, the system can be accessed and used from any system with Internet access and a suitably capable web browser (see fig. 2). Systems like Laptop, Palmtop, Smart Phone, Wearable Computer, which are used by the users of Remoter, can be used without any additional software. Figure 2. Remoter Home Page 3 Motivation Figure 1. Remoter tree-tier Architecture According to Davenport[12], knowledge projects are more likely to succeed when they can take advantage of a broader infrastructure of both technology and organization. Technology infrastructure consist of technologies that are knowledge-oriented. If these tools and the skills to use them are already present, a particular initiative will have an easier time getting off the ground. The above technicians usually access and retrieve documents and software by searching them on different intranet web-sites. Such intranet web-sites contain hundreds of thousands of documents (white papers, technical specification and manual), forums and software about ADSL components, that help technicians in the provisioning and assurance activities. Therefore, because technicians were skilled using intranet web-sites, forums and search engine we chose using a unique web-site with Autonomy T M and Snitz forum facilities. As regards the CCBR chosen has been motivated by two reason. On the one hand, CBR technology allow fast search and access to the knowledge stored in form of cases. This is a competitive factor especially in the Telecommunication field. Take the activities of the help-desk technicians. They usually help on-thefield technicians by reading some technical documents, asking to ex- pert for advice or submit a new trouble tickets to some ADSL experts. In such cases, the time is usually of the essence because the on-the-field technician is on the telephone in real time. In contrast, CBR application require someone to input a series of cases, which represent knowledge expressed as a series of problem characteristics and solution. Then, when a on-the-field technician analyst is presented with a problem, its characteristics can be compared against the set of cases in the application and the closed match is selected. On the other hand, the most successful commercial application of CBR is CCBR. They has been commercially successful in resolving help desk problems. Unlike expert system, which require that rules are well structured with no overlaps, case structure can reflect the fluid thinking that goes in the technicians minds. In fact, help-desk technicians, typically spend much of their time trying to connect new problems with known problems and resolving them with known solutions rather than devising new solutions. They recognize problems partly from the symptoms that are presented to them and partly by asking questions to elicit further information about the problem. As they ask questions and receive answers to the questions, they are able to narrow the search for a solution. CCBR works in a manner similar to that used by an help-desk technicians. Therefore, at the end of a benchmark study, we chose to use egain T M Knowledge. 4 Remoter: user roles and functionalities There are different kinds of user roles and, as a consequence, different functionalities of Remoter system. We have identified three main kinds of user roles, as following: 1. basic user: accesses to Remoter OK but not modify it. She can also send her feedbacks, message and new knowledge record to Remoter which will be validated by the knowledge manager. 2. expert user: accesses to Remoter OK but not modify it. Remoter can also ask her to review some new knowledge record. 3. admin or knowledge manager: monitors, manages and maintains Remoter OK. Because of two groups of users (help desk/back office users, on-thefield users), various features of Remoter system has been developed to ensure a good access and usage of the system by the all users:

3 Search: it provides to the users a search and retrieval engine, which matches the users queries presenting an ordered list of knowledge records and forum messages. The users can tune the number of results that will be displayed and the minimum threshold of relevance. They can also select a check button if they want a brief summary for each search result. After inserting their query, the users can then simply click and follow the systems internal pointers to the selected link. For each selected link, a horizontal window will be open which contains all related knowledge record to that link. Technicians can use this functionality in real-time: for instance, when they have a problem (modem trouble, software trouble) whose they know the solution (installation manual, driver, etc), but not where it is. Diagnosis: it provides to the users the capability for finding out a solution related to diagnosis troubles. This module of Remoter is based on a CCBR approach [10]. It interacts with a user in a conversation to solve a complex query, defined as the set of questions selected and answered by the user during a conversation (see fig. 3). The match between the users answers and the question and answer pairs that make up a particular case, determinates which case scores the highest, and therefore, which resolution or action the search recommends. The technicians can use this functionality in real-time, for instance when they have a router or modem trouble whose they know a set of symptoms but not the solution. Feedback: thanks to this feature, technicians could send to Remoter feedbacks at any time. Let us consider the diagnosis process. Once the operator reaches a the solution, Remoter asks her/him to submit a feedback related to the solution. In the feedback form, the users can insert a textual feedback, upload a file and select a vote for the adequacy of the proposed solution. However, they could forget to fill in the feedback form or they can also close the Web Browser or the connection in order to apply the solution proposed by Remoter before to fill in the feedback form. The connection could close itself because of session ended or a network trouble. Therefore, at every log on, a feedback form is proposed to the user. The operator can choose to fill in or to skip the feedback form. In this way, the users have always the chance to submit their feedbacks even if the connection failed or closed. Moreover, the user can see all feedback related to the selected solution. The user feedbacks are very important because allow the knowledge manager to upgrade and maintain the case base. In fact, the knowledge manager can easily check the user feedbacks attached to all solutions navigating a tree, which represents the case base. In this way the knowledge manager can update the case base, delete obsolete out-of-date knowledge or insert new knowledge. Docs: Docs functionality allows the users to navigate through some knowledge record folder. Document repository is divided by thematic areas (see fig. 4). This functionality is like a file system paradigm because it is based on the assumption the user knows what kind of thematic area the knowledge record belongs to. Moreover, the knowledge record is labelled as a news if it is just inserted in Remoter. The knowledge manager can only insert new knowledge records, and for each them he/she can check the button news and insert a brief summary. Forum: according to Marwick [2], collaboration systems like forum can support the conversion of tacit to explicit knowledge. Online discussion database are potential tool to capture tacit knowledge and to apply it to immediate problems. Thanks to this functionality, the user can post and read messages i.e. can share his/her solution to a particular problem. The archive of the forum becomes a repository of useful knowledge. In Remoter, we realized the forum using Snitz forum 2000, an interactive and freeware discussion environment. News: it allows reading the last news (knowledge records, feedback, messages, software) inserted by the knowledge manager. The users can register themselves to the automatic newsletter checking their interest areas. Remoter will send an to all users subscribed to thematic area where a news has been inserted. Figure 3. Figure 4. Diagnosis functionality Docs functionality 5 Remoter: a system based on OKM Framework Remoter is a KMS, which aims to capture, share and apply the OK developed by skilled telecommunication technicians during their day-to-day activities. Nowadays, Companies like Telecom Italia are

4 increasingly distributed. As a result, it is more difficult to collaborate with ones peer. Therefore, they have recognized that in order to remain competitive, they must efficiently and effectively create, locate, capture and share the technicians knowledge (OK) in an innovative way. This is what we called Operational Knowledge Management (OKM). The OK is converting from simple and explicit procedure to tacit and individual competence (i.e. knowledge based on individual experience developed by exchanging information among people). We claim that OK becomes an innovative factor in a competition only if it becomes explicit and shared. Therefore, we have defined an Operational Knowledge Management Framework (OKMF) that is a local approach to Knowledge Management. The OKMF is defined by three distinct phases [4] [1] [3] (see fig. 5): Figure 5. OKM phases 1. Knowledge Acquisition: is the phase of capturing existing domains knowledge and storing it into a repository (in a structured manner); 2. Knowledge Dissemination: is the phase where the repositorys knowledge is accessed and used by the users; 3. Knowledge Up-grade: is the phase of monitoring and up-grading repositorys knowledge during peoples day-to-day activities (such as interacting with other people and computer systems). From the temporal point of view, we have the Acquisition of knowledge. Then it will be possible to use it, during the Dissemination phase. Afterwards, there should be a cycle of use and upgrade of knowledge. This means that the users will be able to add new knowledge to the repository during the knowledge use as well. In this way we do not need to acquire this new knowledge during a formal phase (Acquisition), but we derive it directly by the usage of the OKMS itself. However, it will be necessary to go back to the Acquisition phase, usually when the knowledge into the repository becomes outof-date or inconsistent. 5.1 Remoter Knowledge Acquisition Phase The Knowledge Acquisition is the phase of capturing existing domains knowledge and storing it into a repository. According to Vargas [9], three parties play an important role in the success of the knowledge acquisition process: the knowledge elicitor, the knowledge programmer and the expert. In our approach, both the knowledge elicitor and programmer are the same person, also called knowledge engineer. Because of different Remoter main functionalities (DOCS, Forum and Diagnosis), we can distinguish three main different kinds of knowledge source to acquire: knowledge records (i.e. technical documents, pieces of technical information, norms, software etc...), messages posted in some existing forum related to ADSL domain and technicians expertise. In order to acquire them, we used the same knowledge acquisition approach. In particular we can subdivide the knowledge acquisition process in four phases: selecting the professionals who qualify as experts, collecting OK, structuring OK according to the different functionalities and validating of the structured OK. As regards the Docs functionality we gathered the knowledge records from the experts, published and structured them according to a file system hierarchy. Instead, as regards to the Forum functionality we gathered some posted messages from existing ADSL forum and published them structuring according to categories of ADSL context. Both Docs hierarchy and Forum categories are validated with the expert users themselves. Finally, we consider the Diagnosis functionality, which is built on CCBR shell. CCBR is a problemsolving paradigm that is able to utilize the specific knowledge of previously experienced, similar problem situation (case) to solve a new problem. A case is a list of question and answer pairs. Therefore, the case base, that is a set of cases, can be thought as a tree. In particular, egain T M Knowledge provides to the knowledge engineer a very user-friendly interface, called egain T M Knowledge Author, which allows building step-by-step the case base. Therefore, starting from the OK sources, we planned the overall case base structure in term of case clusters, and associated links among them. Each case cluster has been identified by the symptoms of the problem to include. Then starting from the symptoms, we have recognized the underlying problems, and for each of them we have listed all possible solutions. Once we had the top (symptom/problem) and bottom (solutions) of the case cluster, we have created a consistent and efficient route between the problem and its solution. This route is composed by a list of question and answer pairs. Finally, in order to get a consistent and efficient list of question and answer pairs we have followed some sample CCBR guidelines [10]. 5.2 Remoter Knowledge Dissemination/Up-grade Phases Most traditional knowledge management systems do not consider a formal knowledge up-grade phase. Especially, they do not relate the Knowledge Dissemination with the Knowledge Up-grade phase. Fischer and Ostwald [5], for example, claim that traditional knowledge management systems are based under assumption that all the knowledge needs are stored into the repository during the knowledge acquisition phase. In fact, once that all domains knowledge is stored into a repository, the users are able to access and retrieve the stored knowledge at least until it becomes out-of-date or inconsistent, such as for a strong innovation technology. In other words, this perspective treats knowledge workers as passive recipients of information. Our OKM perspective is based under assumption that knowledge workers create new knowledge during their work and mainly when they use knowledge of the KMS. Therefore, Remoter allows telecommunication operators to integrate new knowledge at use time. As regards the Diagnosis functionality, the upgrade

5 of case base is managed by knowledge manager. Basic or expert users can send feedbacks related to the solutions proposed during the diagnosis process. Afterwards the knowledge manager can validate or reject feedbacks in order to upgrade the case base (see fig. 6). Figure 7. Knowledge UpGrade process Figure 6. Managing feedback example As regards Forum functionality, the upgrade is managed by a knowledge manager or a moderator. The operators can post messages to the Forum. Then the moderator can edit or delete the new messages posted. Finally, as regards the Docs functionality we developed a knowledge up-grade strategy in which upgrades are reviewed and analyzed by trusted and reliable reviewers. This knowledge Upgrade strategy can be divided in two methods: 1. Direct Knowledge Up-grade: the knowledge manager can only add new knowledge directly; 2. Indirect Knowledge Up-grade: the new knowledge record (candidate) is added by the knowledge manager after it is reviewed by a set of expert users. In Remoter there are three different kinds of user: basic user, admin or knowledge manager and expert user. Each user can be an expert in one or more thematic areas of ADSL context. In particular, this methods can be divided into steps shown in fig. 7. Once a user uploaded a candidate knowledge record, he/she must check the thematic areas whose the knowledge record belong to. Afterwards the candidate will be send to all expert users belong to thematic areas of the candidate. The expert users can voice a vote, attach a comment and send an to the user who sent the new candidate knowledge record. At any moment, the knowledge manager can check the grand total of expert users who reviewed the candidate and the average of its votes. In this way the knowledge manager can publish the new knowledge record or not. 6 Related work Turning to related work, Smith and Farquhar [13] describe an approach to Knowledge Management which combines different areas and technologies. They describe their experience at the Schlumberger company with developing a KMS, called InTouch. InTouch allows sharing and applying knowledge among different Schlumberger s CoPs which are engaged in what they called field activities often on remote seismic vessels or drilling rigs to find and develop oil and gas reservoirs. As our technicians from the ADSL context, Schlumberger s CoPs are directly involved in the field activities and share a set of procedures or norms need to do their jobs. Both approaches, InTouch and our emphasize the importance of upgrading the knowledge hub (i.e. the knowledge repository) to memebers of CoPs (i.e. technicians) during their normal work. However, InTouch does not support technologies which represent knowledge in a structured manner like CBR. On the other hand, InTouch offers knowledge management functionality that are no part of our system like expertise directory, best practices repository. The substantial list of deployed applications of CCBR for Knowledge Management include, for example, a knowledge management system for PC troubleshooting [15]. As our approach, the author emphasize the importance of CCBR technique of supporting the knowledge management activities, including knowledge capture, knowledge development, knowledge sharing and knowledge utilization. When compared to our approach this system lack the upgrade level that is partially provided in our approach by using the feedback mechanism. On the other hand, the author combined ontology with the CCBR techniques to ensure the effective communication among users and system. Another application that uses CCBR technique was implemented at NEC [14]. NEC is using knowledge stored in a case library to reach unprecedented levels in quality control and marketing of its products. Alike our approach an important issue they have addresses refers to maintenance of case bases that used to be update manually. The obvious problems arising from manual update were overcome with the creation of FAQ case bases automatically, allowing creation and understanding to be implemented without human intervention. The system, called SignFinder, detects variation in the case bases generated automatically from customer calls. When they detect variations on the content of typical customers requests, they can discover knowledge about defects on their products. Besides, the automatic update of case library allows the timely sharing of new

6 defects solving the knowledge sharing obstacle they face due to the magnitude of their organization. 7 Future work Due to the complexity of the ADSL domain each technicians has his/her own vocabulary to express the operational knowledge. In the context of knowledge sharing, as Remoter, to have different vocabulary could be a drawback. In recent years the development of ontologies has moved from the realm of Artificial Intelligence laboratories to the desktop of domain experts. Ontologies have become common on the World-Wide Web. Many languages as Resource Description Framework (RDF), Darpa Agent Markup Language and Ontology Inference Layer (DAML+OIL), Karlsrurhe Ontology (KAON) and OWL (Ontology Web Layer) were developed for encoding knowledge on the Web to make it computer processable. Therefore, we are developing an ADSL ontology in order to share common understanding of the operational knowledge among people. We are using Text-to- Onto, an open-source ontology management infrastructure, in order to create an ADSL ontology [11]. It facilitates ontology engineering by discovering taxonomical and non-taxonomical conceptual relationship between concepts from texts. The aims of the ADSL ontology are on the one hand to have an ontological commitments for telecommunication technicians and on the other hand to improve the Knowledge Up-grade phase with regard the Diagnosis context. The Knowledge Up-grade phase for the diagnosis functionality is related to the feedbacks which are submitted by telecommunication operators. This feedbacks are related to the solutions proposed by Remoter. Moreover, Remoter provide an ASP window to the knowledge manager which represents the case base with those feedbacks. Thanks to this ASP window and to the knowledge author module provided by egain, the knowledge manager may upgrade the case base. We would, for instance, using ADSL ontology, be able to suggest the appropriate concepts for feedback to the telecommunication operators in order to automatize the up-grading of the case base. Future issues that will be also explored include using of the agent technologies in order to capture users profiles, to automatize the question and answer flows with regard the diagnosis functionality, to have a distribute knowledge base (document repository, case base and messages) and finally to build an Expertise Location System which have the goal of suggesting the names of person who have knowledge in a particular area [2]. 8 Conclusions and experiments Remoter is an Operational Knowledge Management System (OKMS) that enables telecommunication technicians from ADSL context to share, capture and apply their collective operational knowledge (OK) to take optimal decision in real time. In particular, two groups of users have been identified: help desk/back office users, on-the-filed users. Because of two groups of users, various features of REMOTER system has been developed to ensure a good access and usage of the system by all the users: Docs, Diagnosis, Forum, Feedback, News, Search. However, the proposed system is a three tier web-based application. Therefore being a web service, the system can be accessed and used from any system with Internet access and a suitably web browser. For the Remoter development an Operational Knowledge Management Framework (OKMF) has been followed [1]. The OKMF is based of three main phases: Knowledge Acquisition, Knowledge Dissemination and Knowledge Up-grade. The Knowledge Acquisition campaign has been conducted on the experts of the ADSL context. Because of different Remoter functionalities, OK has been acquired modelled and then implemented into different structures (knowledge record, case base, messages). One of the main features of our OKMF is the relation between the knowledge dissemination and the knowledge up-grade phases. In fact, our OKM perspective is based under assumption that knowledge workers create new knowledge during their work and mainly when they use knowledge of the OKMS. Therefore, Remoter allows telecommunication operators to integrate new knowledge at use time. Finally, besides Diagnosis and Forum functionalities, we have introduced a knowledge up-grade strategy with regard to the Docs functionality in which upgrades are reviewed and analyzed by trusted and reliable reviewers. Initial assessment of Remoter has been very positive. In particular Remoter has been field proved for the last six months of the 2003 in two Telecom territorial areas (TA) of Italy. Significant saving in the energy, time and number of phone calls from on-the-field to help/desk technicians has been realized with the help of Remoter. Each TA had about 30 technicians (one sixth of them were help/desk technicians) which used Remoter. At the beginning of the trials, Remoter contained about 50 cases, 400 knowledge records and 60 messages posted. After six months, the OK in Remoter has been extended to incorporate more than 100 cases, 700 knowledge records and 250 messages posted. REFERENCES [1] G. Valente and A. Rigallo, An Innovative Approach for Managing Competence: An Operational Knowledge Management Framework, , Proc. of 7th International Conference on Knowledge-Based Intelligent Information and Engineering Systems, LNCS 2773, Springer- Verlag, (2003). [2] A.D. Marwick, Knowledge Management Technology, IBM Systems Journal, , 40(4), (2001). [3] G. Valente and A. Rigallo, Operational Knowledge Management: a way to manage competence, , Proc. of the International Conference of Information and Knowledge Engineering, CSREA Press, (2002). [4] N. Shadbolt and N. Milton and H.Cottam and M.Hammersley, Towards a knowledge technology for knowledge management, Int. J. of Human- Computer Studies, , 51, (1999). [5] G. Fisher and J. Ostwald, Knowledge Management: Problems, Promise, Realities and Challenges, IEEE Intelligent System, 60-72, January/February, (2001). [6] Egain WEB Site, [7] Autonomy WEB Site, [8] Snitz WEB Site, [9] A.G. Vargas and F.C. Hadipriono, Knowledge Acquisition and Knowledge Structure for Safety First Expert System, Proc. of 31th ASC Conference, , (1995). [10] D.W. Aha and L.A Breslow and H.Munoz-Avila, Conversational Case- Based Reasoning, Applied Intelligence, 9-32, 14, Kluwer Academic, (2001). [11] S.Staab and A.Maedche, Semi-automatic engineering of ontologies from text, Proc. of 14th International Conference on Software and Knowledge Engineering, (2000) [12] T.H. Davenport and L. Prusak, Working Knowledge, Harvard Business School Press, Boston, (1998) [13] R.G. Smith and A. Farquhar, The Road Ahead for Knowledge Management, AI Magazine, 17-40, Winter, (2000) [14] H. Shimazu and D. Kusui, Real-Time Creation of Frequently Asked Questions, , Proc. of 4th International Conference on Case- Based Reasoning, LNCS 2080, Springer-Verlag, (2001). [15] S.L. Wang AND H.Hsu, A Web-based CBR knowledge management system for PC troubleshooting, The International Journal of Advanced Manufacturing Technology, , 23(7-8), (2004)