From rule-based to probabilistic expert systems, IoT, IoMe, Industry 4.0 and intelligent agents: HSC Enterprise Computing Intelligent Systems
“Research factors that have allowed expert systems to advance from rule-based systems to integrate probability, including processing power, data availability and access, hardware costs, neural networks and webometrics; describe how people's changing needs have shaped the proliferation of expert systems, including the Internet of Things (IoT), Internet of Me (IoMe) and Industry 4.0; describe the operation of simplistic and complex intelligent agents used by search engines, including predictive search strings and the use of voice assistants to initiate interaction”
Expert systems moved from fixed rules to probabilistic reasoning because of greater processing power, abundant data, cheaper hardware, neural networks and webometrics. Demand for connected devices (IoT), personalisation (IoMe) and smart factories (Industry 4.0) spread them everywhere. Search engines use simple agents for fixed responses and complex agents for predictive search and voice assistants that learn and plan.
Jump to a section
What this dot point is asking
This content covers how intelligent systems evolved and spread. You need to explain the factors that moved expert systems from fixed rules to probability, how changing needs (IoT, IoMe, Industry 4.0) drove their spread, and how intelligent agents in search engines operate.
The answer
From rule-based to probabilistic systems
Early expert systems (1970s and 1980s) used hand-written IF-THEN rules. They struggled with uncertainty, could not learn, and became hard to maintain as rules multiplied. Modern systems integrate probability: they estimate how likely each conclusion is, often learning those likelihoods from data.
Factors that made this possible:
- Processing power: fast multi-core CPUs, GPUs and cloud computing can run probabilistic models and train neural networks over huge datasets.
- Data availability and access: digital records, the web, sensors and open datasets provide the examples needed to estimate probabilities.
- Hardware costs: cheaper processors, memory and storage put these capabilities into phones, cars and appliances.
- Neural networks: layers of connected nodes learn complex patterns (images, speech, language) that rules cannot capture, and output probabilities.
- Webometrics: quantitative analysis of the web (links, citations, clicks) produced both vast data and techniques such as link-based ranking that underpin search and recommendation.
Changing needs and the spread of expert systems
- Internet of Things (IoT): billions of connected devices (smart meters, thermostats, industrial sensors) need built-in decision-making, so expert-system logic is embedded everywhere.
- Internet of Me (IoMe): people expect technology organised around them: personalised recommendations, health advice from wearables, adaptive learning apps.
- Industry 4.0: smart factories combine cyber-physical systems, IoT, cloud and AI for predictive maintenance, automated quality control and flexible production.
Intelligent agents in search engines
An intelligent agent perceives its environment and acts to achieve goals.
- Simplistic (simple reflex) agents follow fixed condition-action rules on the current input (spelling correction that replaces a common typo).
- Complex agents keep a model of the world and user, learn from history, and plan (ranking results by predicted relevance, personalising by location and past behaviour).
Predictive search strings: as you type, the agent predicts completions from popular queries, trends, location and your history, updating with every keystroke.
Voice assistants let users initiate interaction by speaking a wake word. The agent converts speech to text, interprets intent with natural language processing, searches or acts (sets a timer, answers a question), and replies with synthesised speech. They are complex agents that learn from interactions and context.
A user asks a voice assistant, "What's the weather in Orange tomorrow?"
- Wake word detected on the device; audio is sent to the provider's servers.
- Speech recognition converts audio to text.
- Natural language understanding identifies the intent (weather forecast), location (Orange, NSW, using the user's region to resolve which Orange) and date (tomorrow).
- Action: the agent queries a weather service.
- Response: it generates a spoken reply and may learn that the user often asks about Orange, suggesting it in future.
- Saying rule-based systems have disappeared
- Rules are still used where decisions must be explainable; many modern systems combine rules and probability.
- Confusing IoT with IoMe
- IoT is about connected things; IoMe is about personalising technology around the individual.
- Describing voice assistants as simple agents
- They use complex language models and learn from context.
Practice questions
Original practice questions graded from foundation to exam level, each with a full worked solution. Try them before revealing the solution.
foundation3 marksExplain how predictive search strings work in a search engine.Show worked solution →
As the user types, the search engine's agent sends each keystroke to its servers and predicts likely completions. It ranks suggestions using how often other people searched similar strings, trending searches, the user's location, language and (if signed in) their own search history. The suggestions update with each character, saving typing and guiding the user to common queries.
Marking guide: 1 mark for real-time prediction as the user types, 1 mark for data sources used, 1 mark for ranking or personalisation.
core4 marksResearch factors that allowed expert systems to move from rule-based reasoning to probabilistic reasoning. Explain two factors.Show worked solution →
Data availability and access. The web, sensors and digital records produce huge datasets, so systems can estimate probabilities (how often a symptom goes with a disease) from real cases instead of relying only on rules written by experts.
Processing power and hardware costs. Faster, cheaper processors, GPUs and cloud computing make it possible to train neural networks and run probabilistic calculations over millions of examples quickly, which was impractical when early expert systems were built.
(Neural networks and webometrics are also valid factors.)
Marking guide: 2 marks per factor explained with its effect.
exam6 marksDescribe how people's changing needs, through the IoT, the Internet of Me and Industry 4.0, have shaped the proliferation of expert systems. Use examples.Show worked solution →
- Internet of Things
- People expect everyday devices to be smart and connected. Smart thermostats, security cameras and appliances embed rule and probability-based reasoning to act automatically (lowering heating when nobody is home), so expert systems now run inside millions of devices rather than only on specialist computers.
- Internet of Me
- People want services tailored to them. Wearables and phones collect personal data, and systems use it to give individual advice (a fitness app adjusting a training plan, a banking app flagging unusual spending). This demand for personalisation has spread expert-system reasoning into consumer apps.
- Industry 4.0
- Businesses need efficiency and flexibility. Smart factories use sensors and expert systems to monitor machines, predict failures and adjust production automatically, and digital twins simulate changes before they are made.
- Overall
- Expert systems have moved from rare, standalone tools to embedded, everyday components because people and enterprises now expect automation, personalisation and real-time responsiveness.
Marking guide: 2 marks each for IoT, IoMe and Industry 4.0 with examples (6 marks), with credit for an overall link to changing needs.