IBDP Computer Science A3.4 Alternative databases and data warehouses HL Paper 1 - New Syllabus

(a)
For the correct answer:

• A network database has a web-like structure of nodes in which data is linked to other data points.
• It can support multiple parent records and multiple child records, allowing many-to-many relationships between entities.
• It allows flexible navigation through natural relationships between connected records.

Example: A criminal could be linked to several crime records, while each crime could also be linked to several criminals.

(b)
For the correct answer:

Advantages of an object-oriented database:

• Object-oriented databases can efficiently represent and manage complex data structures and relationships, such as multimedia data, polygons and points, rather than only simple values such as integers and strings.
• They use features similar to object-oriented programming, including inheritance, encapsulation and polymorphism. These features support code reusability and can simplify application development.
• They can provide improved performance because program objects can be stored directly. A relational database may need to flatten objects before storing them and recreate them when retrieving them, creating an object-relational impedance.
• They can model real-world objects and relationships effectively. The functionality of objects can be extended through inheritance and polymorphism.

Disadvantages compared with relational databases:

• Relational databases are more widely used and accessible, so there is greater support available for the relational model.
• The availability of many relational database vendors reduces the risk of vendor lock-in.
• Relational databases generally have greater compatibility and interoperability with other systems.
• Relational databases are easier to design, modify and extend, whereas object-oriented databases can be more complex to modify and extend.
• Relational databases have standard query tools and report generators.
• Relational databases can provide greater integrity and lower redundancy through constraints such as referential integrity and domain integrity.
• Relational databases can provide better performance in heavy multi-user environments because their data model can handle concurrency effectively.
• Relational databases are generally better suited to large datasets and complex queries.

Conclusion: An object-oriented database would be particularly useful where Europol needs to represent complex objects and relationships directly, but a relational database may be preferable where compatibility, standard query tools, mature support and large-scale data management are more important.

(c)
For the correct answer:

• A data warehouse centralizes data from multiple sources, allowing Europol to analyse crime information without carrying out separate searches on different sources.
• Data can be cleaned and standardized through ETL processes, improving the consistency of the data available for analysis.
• A data warehouse can store much larger amounts of historical data than a typical operational database, increasing the likelihood of finding patterns and identifying anomalies.
• Data is time-stamped, allowing historical intelligence to be analysed and trends to be identified.
• Advanced OLAP tools can support techniques such as data mining, data matching and link analysis, as well as classification, clustering and association analysis.

Explanation: A data warehouse is designed primarily for analysis rather than day-to-day transaction processing. This makes it useful for combining large quantities of historical crime data and identifying meaningful patterns.

(d)
For the correct answer:

• Predictive modelling can be used to forecast likely future crimes by analysing factors such as demographics, socio-economic indicators and environmental factors.
• The results can help Europol allocate limited resources more effectively and optimize resource deployment.
• Predictive models can update as the underlying data changes, allowing predictions to reflect new information.
• Predictive modelling can focus on information-bearing patterns while reducing the influence of random patterns, errors, outliers or other noise.
• It can identify crime patterns associated with particular times or events, allowing law enforcement agencies to prepare appropriate strategies.

Example: Historical crime data could be used to identify patterns showing that certain types of crime are more likely to occur in particular locations or during particular periods. Europol could then use these predictions to plan resources and investigative strategies.

(e)
For the correct answer:

• Link analysis identifies relationships between large quantities of crime data.
• It can examine relationships between individuals, locations, organizations, objects and events.
• It can associate data from multiple databases to identify connections that may not be obvious when the datasets are examined separately.
• Link analysis can discover new patterns involving individuals, locations, organizations, objects or events.
• Data matching techniques can be used to identify similarities between datasets and establish possible links.
• It can identify anomalies where known patterns are violated, which may indicate fraudulent or criminal activity requiring further investigation.

Example: Europol could analyse phone records, financial transactions and information about known criminals. If several individuals repeatedly communicate with each other and their financial transactions are linked to the same fraudulent activities, link analysis could reveal a possible criminal network.

Similarly, patterns in credit-card transactions could be linked to identify fraudulent activity or connections between different criminal investigations.