The ISTN 731 chatroom is a discussion forum on Hipchat hosted by Craig
Blewett and Rosemary Quilling, lecturers of special topics, a second semester module for ISTN honours 2014 students of the University of KwaZulu-Natal, Westville. Few weeks before then, lecturers Craig and Rose gave the class an assignment which involves using twitter as a tool for knowledge management, personally, I have had a twitter account since 2012 and had never considered it a tool for knowledge management. On Tuesday, 16th September, 2014, we held a chatroom session about the use of microblogging for knowledge management, using twitter. Alan Kyeswa opined that the good thing about Twitter is the possibility of sharing your thoughts and knowledge across an unlimited audience who themselves are twitters and bloggers; however, the limitation is the restriction of text character to only 140 characters. While most students agreed that microblogging is an effective tool for knowledge management, Dhevesh Parumasur thought that microblogging may fail as a tool for knowledge management in organizations because employees use their knowledge to get ahead of each other for promotions and other advantages and may not be willing to share their knowledge, therefore, formal knowledge management systems may fail. He further said that if companies use microblogging sites such as twitter, integrity might be an issue as anyone would be able to get knowledge which should be confidential and could give a competitive advantage to the company by just following the company but that taking a look at enterprise solution in an organization could increase the integrity of microblogging and confidentiality of information. Another challenge which was unanimously agreed upon is that there are streams of information on microblogging sites such as twitter and it could be very hard to mine out relevant data when needed. The solution to this problem is to make use of the # tag when posting information, however, the challenge with using # tag remains that some individuals may not tag their post and tags when used, are user curated.
KEY BENEFITS OF USING A MICROBLOG FOR KNOWLEDGE MANAGEMENT Micro-blogs allow users to exchange small elements of content such as short sentences, individual images, or video links. 1. Ambient Awareness: In combination, different tweets sent out over time can paint a very accurate picture of a persons activities. Several tweets together can generate a strong feeling of closeness and intimacy. Due to ambient awareness, applications such as Twitter result in relatively high levels of social presence, defined as the acoustic, visual, and physical contact that can be achieved between two individuals (Short, Williams, & Christie, 1976); and media richness, defined as the amount of information that can be transmitted in a given time interval (Daft & Lengel, 1986). 2. Push-Push-Pull Communication: Following implies that one authors tweets are automatically pushed onto the Twitter main page of all followers. Following another user creates an additional element of convenience as it reduces the effort associated with accessing this information. In some cases, the receiver of the message might find the news so interesting and intriguing that they decide to give it an additional push by re-tweeting it to their own followers. Specifically, when the initial message has been sent out by a company, this transformation of a commercial message into buzz can substantially increase its impact and credibility (Katz & Lazarsfeld, 1955). Once the message has been pushed and pushed again through the whole network, it may motivate some user to go out and pull additional information on the subject from other sources. 3. Virtual Exhibitionism and Voyeurism: Because tweets are public by default, unless otherwise requested, any message sent through twitter automatically becomes public knowledge within minutes of its publication. This creates the perfect environment for virtual exhibitionism and voyeurism, these two factors represent the third reason behind growth of microblogging applications PROBLEMS OF USING MICROBLOG FOR KNOWLEDGE MANAGEMENT 1. Security: Some informants reported that they might share even more work related updates and would like to be aware of similar things from co-workers, if Twitter could be a safe place to post inside the companys firewall. Although people hesitate to mention project-or client-specific information on a public feed, some still worried whether it would be a safe place to explicitly discuss business-sensitive information and care need to be taken about what is said as one never knows in case someone from outside the company spies at the conversation. 2. Privacy: In Twitter, subscribing to ones updates is open without permission approval, and the system sends a users updates to all his/her subscribers. An employee may have concerns about what to update if his boss is in this subscription list. A manager may hesitate to update because he may not want all his team members to know what he has been doing. 3. Integration: People dont only use twitter for knowledge management, some use it to keep up to date with their friends and family and even business partners. So many people find it difficult to integrate these two functions together. 4. Filtering and grouping: Another problem with the use of microblogs for knowledge management is that some people could end up following too many people; this could result in cognitive overload; where users have a hard time monitoring a large amount of people and following their updates. I recently had to unfollow a pastor on twitter because he overloads my page with so much sermons and motivation I could hardly see anything else Experts have defined micro-blog as a group of Internet-based applications that build on the ideological foundations of Web 2.0 and that allow the creation and exchange of user generated content (Kaplan & Haenlein, 2010, p.61) keeping users in contact with each other without necessarily incurring excessive cost (Bughin, Chui &, Miller 2009) this enhances communication with a wide array of audience and consequently yield increase in collaboration and overall productivity in organizations (Hoong, Ming, Aripin & Aun, 2012)
REFERENCES 1. Bughin J., Chui M.,Miller A., (2009), How companies are benefiting from Web 2.0, McKinsey Global Survey Results. 2. Daft, R. L., & Lengel, R. H. (1986). Organizational information requirements, media richness, and structural design. Management Science, 32(5), 554571. 3. Hoong A.L. S., Ming L. T., Aripin R., Aun J. L. R.,(2012), A comparison study on the use of knowledge management systems and enterprise microblogging systems for organizational knowledge sharing, 2nd international conference on management (2nd ICM 2012) proceeding. 4. Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of social media Business Horizons, 53(1), 5968. 5. Katz, E., & Lazarsfeld, P. F. (1955). Personal influence: The part played by people in the flow of mass communications. Glencoe, IL: The Free Press. 6. Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications. Hoboken, NJ: John Wiley & Sons, Ltd.
EFFECTIVENESS OF MINING A MICROBLOG FOR KNOWLEDGE MANAGEMENT Good things they say do not come easy, so is data mining. Just like gold is mined out, refined and turned into asset, there is a need to dig out data from the sea of knowledge that exist in the world of the web 2.0. As microblog grow more popular, microblog services have become information provider on the web scale ( Zhang & Sun, 2012). Needless to say, microblog has become a popular social networking service in recent time because of the following reasons: Ease of use and convenience Real-time platform to update personal status Possibility to publish microblog over multiple delivery channels (web interface, cell phones, etc.) Hang & Lerman (2010) divided microblogs into three categories: Broadcast Conversations or Retweet messages Whatever the category however, how to effectively dig out latent topics and internal semantic structure from large scale data is a very challenging issue (Zhang & Sun, 2012). More than ever before, organizations and business organizations have realized that information is as much asset as the financial and infrastructural assets, if not more, and that the overall business strategy and capability must be well in alignment with the IT strategy and capabilities to produce value for all stakeholders. Most of them have created an online presence and taken onto social media and microblogging resulting in streams of information to form a robust body of knowledge on the web 2.0. Though microblogs contain structured information, they are short text (140 characters) and carry limited information. This makes it very difficult to mine out relevant topics. To solve this problem, the microblog tag feature, for example the #tag feature on twitter has become of immense use (Gunther & Krasnova 2009). This has made it easier to find relevant information when mining a microblog as against the traditional knowledge mining. For example, I made use of the #tag feature to mine out information about post by the 2014 ISTN honours class on twitter, using #ukzn2014, and in seconds I had information regarding posts of ISTN 2014 honours students which has been tagged as below:
Zhang & Sun (2012) also proposed the use of a novel probabilistic generative model based on LDA called MB-LDA. This takes contactor relevance and document relevance into consideration to improve topic mining in microblog. COMPARISM BETWEEN MICROBLOG MINING AND TRADITIONAL MINING FOR KNOWLEDGE MANAGEMENT The traditional mining for knowledge management is usually in the form of plain text and there are quite a few methods, models or algorithms used. The clustering methods are early solution which transfer unstructured text data to vectors by Vector Space Model (VSM) and do clustering with traditional methods like K-means (Xu & Wunsch, 2005). Clustering results are often considered sharing the same topic respectively. However, the major disadvantage of clustering methods is that many of these algorithms depend on distance functions for the pairwise distance measurements, which are difficult to define in large scale corpora; besides, there lies no semantic information in clustering results, which need further analysis to extract topics Dimensionality reduction method like Latent Semantic Analysis (LSA) was introduced to text mining by Deerwester et al (1990). By assuming that words close in meaning will occur close together in text, LSA constructs a term- document matrix in the popular tf-idf scheme (Salton & McGill, 1983) and use singular value decomposition to capture its latent semantic in the concept space. Although LSA can extract topics from corpora and find relations between terms in a semantic way, the limitation is that the result of SVD is less interpretable and LSA itself cannot capture polysemy. LSA is not suitable to large scale text mining due to its high computing cost Probabilistic topic models such as Latent Dirichlet Allocation (LDA) was introduced to text modelling by Blei et al (1998). LDA is also a method to recover the latent topic structure, which extends PLSA ( Hofmann,1999) by defining a complete probabilistic generative model. The intuition behind LDA is that documents exhibit multiple topics which are represented by distributions over words. In the framework of LDA, words of documents are the observed variables while topic structures are the hidden variables. Through probabilistic inference for LDA, the hidden variables are inferred and topics can be discovered from the corpus (Griffiths & Steyvers, 2006) Different from plain text, microblog has its special symbols (i.e. @ and RT) to characterize the relation between microblogs: @ indicates the contactor relevance relation of microblogs and RT indicates the document relevance relation of microblogs. Contactor relevance relation of microblogs refers to that conversation message and its contactor (@) have latent semantic relationships. In general, messages with the same contactor usually share the related topics. This phenomenon is very common in conversation microblogs. Document relevance relation of microblogs refers to that comment on retweet message and its original content have latent semantic relationships. In general, the comment part and original part share the related topics. This phenomenon is very common in retweet microblogs. In lay terms from my experience while mining information traditionally form sites such as Google for my research proposal; I observed the following difference between plain text or traditional data mining and microblog mining: S/N Microblog Mining Traditional Mining 1.
Rich in information from diverse viewpoints, opinions and perceptions because of comments from different people on the same topic or tweet Information is usually a monopoly of one mans viewpoint, opinion or perception
2. Information is usually on trending issues and are up-to-date
Usually repositions of archived knowledge which in most cases are not are up-to-date as the microblog. 3. Information is short, concise and straight to the point because of the character limit to 140
Author can perambulate as much as he deems appropriate, therefore, lots of sieving needs to be done to get the actual points needed
4 Microblogs could contain shortened URL links to other sites containing relevant information, the site could shed light on doubt and questions that the researcher could have and even open up the possibility to contact the author for further clarification Information could be unclear and confusing, even then, there is hardly any provision to contact author for clarification
From the points raised above, I can safely conclude that it is easier and more rewarding to make use of microblog mining rather than the traditional means of mining for knowledge management.
REFERENCES 1. Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. The Journal of Machine Learning Research, 3: 9931022. 2. Deerwester, S., Dumais, S., & T. Landauer, T., et al (1990). Indexing by latent semantic analysis. Journal of the American Society of Information Science, 41(6): 391407. 3. Griffiths, T. & Steyvers, M. (2006) Probabilistic topic models. Latent Semantic Analysis: A Road to Meaning. Hillsdale, NJ: Laurence Erlbaum. 4. Hofmann, T. (1999). Probabilistic latent semantic indexing. In Proceedings of the 22nd annual International ACM SIGIR Conference on Research and development in information retrieval, 50-57 5. Kang, J.H., Lerman, K., & Plangprasopchok, A. (2010). Analysing Microblogs with Affinity Propagation. In Proceedings of the 1st KDD workshop on Social Media Analytic, 67-70. 6. Landauer, T.K., Foltz, P. W., & Laham, D. (1998). Introduction to Latent Semantic Analysis, Discourse Processes, 25: 259-284 7. Salton, G., & McGill, M. (1983). Introduction to Modern Information Retrieval. New York: McGraw-Hill. 8. Xu, R., & Wunsch, D. (2005) Survey of clustering algorithms. IEEE Transactions on Neural Networks, 16(3): 645678. 9. Zhang, C. & Sun, J. (2012). Large scale microblog mining using distributed MB- LDA. International World Wide Web Conference Committee (IW3C2).
Unit-4object Segmentation Regression Vs Segmentation Supervised and Unsupervised Learning Tree Building Regression Classification Overfitting Pruning and Complexity Multiple Decision Trees