BUSINESS INTELLIGENCE
This is the practice of extracting useful knowledge from the voluminous data available on the web. Web usage mining is the basis of web site management of e-commerce sites through the creation of adaptive web sites, business and support systems as well as personalization and network traffic. The www platform has grown into an influential gateway for the management and conduction of business operation. The www can be classified as content, structure and usage mining. Web servers accumulate a lot of data whenever an interaction is initiated by the user. Analysis of the web logs is important to the user and the business entity to efficiently manage the activities related to e-business, e-services and e-education.
The data obtained in a mining process need to be analyzed, transformed and aggregated at different levels of abstraction. Thus, data content is one value added feature utilized by e-commerce sites to convey a relationship with the user. Data content is comprised of textual and graphic representation sourced from HTML/XML pages and scripts. The content data involves the semantic or structural meta-data that are embedded within pages sites, attributes descriptive keywords among others.
Next value added feature has to do with the user data or the operational databases for the site including the user profile information among others. Features such as demographic information about registered users, user ratings and history details such as purchase, visitor visits.
Structured data highlights the views of the designers about content organization in the website. For instance HTML and XML documents may be represented as tree structures over the tags page while hyperlinks are automatically generated by the site map.
There are three elements of web usage data processing; data fusion and cleaning, Page view verification and user identification. Data fusion is the merging of log files originating from diverse web servers and applications to through a global synchronization process. Merging is done through embedded session ids, heuristic methods and inter site web mining. Identification of the page views is dependent on the intra-page structure of the site, page contents and the underlying site domain contents. It is necessary to distinguish among different users through the user activity records in order to determine the frequency of logging activities of an individual user.
DATABASE
A database is a structured collection of records of data stored in a computer system. The data is arranged in a database model to achieve the structured criteria. Mostly data in a data base are arranged in a relational model among other models such as hierarchical and network. Database uses an Online Transactional Processing but can still be utilized for other purposes such as data warehousing. The tables in a database are complex since they are normalized in RDMS to reduce redundancy.
A data warehouse, on the other hand, represents a\ repository of an organizations data stored electronically. It is primarily designed to permit the reporting and analysis of data. In a data house, the mechanism for the retrieval, analysis, extraction, transformation and loading of the data are significantly essential. The volume of data in a warehouse is usually large.
A warehouse uses an Online Analytical Processing OLAP model that reads the historical data for the user. The tables are easy since they are represented in a de-normalized manner thus reducing the response time for querying. Data modeling techniques are utilized in a warehouse.
A data warehouse is beneficial to end users because it represents a subject-oriented, integrated, non-volatile and time varying collection of data that are meaningful for decision making purposes. Querying and analysis of data sourced from multiple sources is represented in a multidimensional data model thus providing an easy mechanism for drawing conclusions.
Different architectures of warehouses are preferred because of the functional capabilities it accords an institution. For instance, there are different architectural models of warehouses including;
- Data mart architecture DBA
- Enterprise Data Warehouse EDW
- Federated architecture FED
The choice of these architectures is dependent on a number of factors;
Organizations with high interdependency are more likely to implement the EDW and IDM architectures. This is because of the dynamic environments attributed to information processing. Businesses operating in uncertain and turbulent environments envisage the greater urgency in of information systems to provide up to date information. Other factors such as resource constraints, task routineness and perceived ability of the IT staff determine the choice of these architectures.
Organizations that envisage high urgency are more likely to adopt to adopt IDM and EDW and DBA architectures. EDW is better with unstructured non-routine tasks and integrated levels of detail from different data sources.
Resource constraints
The type of architecture adopted is dependent on the resource capabilities of the company, and the information processing capacity envisaged. Different architectures present different information processing capacities thus the choice of a warehouse is dependent on the availability of slack resources. When an organization is face with resource constraints, it chooses architecture that requires little resource allocation.
Ability of the organizations IT staff
The perceived ability of the in-house IT staff in an organization determines the choice of the warehouse architecture. EDW has more pronounced knowledge barriers and its implementation require higher knowledge of the technical skills as compared to IDM and DBA.
EXPERT SYSTEMS
Limitations
Common sense
Typical expert systems cannot generalize through analogy to reason in the same manner that people do. They are, therefore, incapacitated in reasoning about situations the same way people do.
Human beings exhibit common sense, in addition to a great deal, of technical knowledge. However expert systems lack this functionality and the way common sense can be instilled in these systems have not been found.
Creativity
Human experts are excessively creative and can respond to unusual situations and work around it to find a solution. This feature is not available in expert systems as they cannot collect knowledge and interpret into meaningful rules. Developers have not found a way to collect knowledge and refine it into rules that can be used to guide their operations.
Learning
Expert system lacks the automatic mechanism of adapting to new environments and must be automatically updated. The lack of case-based reasoning and neural networks in exert systems require them to be explicitly and constantly updated. They use computational engines incapable of reasoning, therefore, resulting in poor work and abandonment of projects. Case based reasoning and neural networks in human are powerful methods of learning and lack of this functionality makes expert systems lack behind human experts.
Sensory experience
Human experts have the capability to sense a wide range of experiences. Expert systems are, however, dependent on symbolic inputs. The use of variable facts through detection of contradictions and production of explanations make its use more understandable to developers but, such systems are less clear to the user’s, and less reliable thus omitting explanations of their reasoning. Thus, expert systems are not useful in the management of highly sophisticated sensory inputs. Problems presented in multi-dimensional dimensions cannot be effectively be handled by expert systems.
Non-human machines
Expert systems are non-humans machine experts that sometimes do not respond to various situations effectively the way humans would do. The lack of human-self awareness is what makes them less effective. The analysis tools utilized by expert systems lack the needed self-analysis tools.
References
Bidgoli, H. (2004). The internet encyclopedia:. John Wiley & Sons.
Dean, T. (2009). Network+ Guide to Networks. Cengage Learning.