6.8 Data Source Verification Tasks Suitable for Humans

Introduction

The textbook explains that data source verification tasks can vary widely in terms of complexity and the suitability of humans versus computing devices for completing them. The choice between humans and computing devices depends on factors such as the task's nature, accuracy requirements, speed, and the availability of resources. In simple words, some data verification tasks are better done by humans, while others are better done by computers. Understanding which tasks suit which is important for efficient and accurate work.

Definition

Data source verification tasks are activities that involve checking, confirming, or validating the accuracy, credibility, and reliability of data and its sources. Tasks Suitable for Humans are verification tasks that are best performed by human beings because they require subjective judgment, complex decision-making, handling ambiguity, or personal judgments.

Main Idea

The main idea is that humans and computers have different strengths. Some tasks are better suited for humans, while others are better suited for computing devices. The textbook provides a breakdown of these differences. The textbook states: "The choice between humans and computing devices depends on factors such as the task's nature, accuracy requirements, speed, and the availability of resources."

Core Concept

Tasks Suitable for Humans The textbook lists four types of tasks that are suitable for humans: Subjective Judgments1. Complex Decision-Making2. Handling Ambiguity3. Personal Judgments4. Figure 6.15 (Page 176) shows tasks suitable for humans. Tasks Suitable for Computers The textbook lists six types of tasks that are suitable for computers:

Speed and Reliability1. Processing Large Amounts of Data2. Statistical Analysis3. Pattern Recognition4. Speed and Consistency5. Automated Verification6. Figure 6.16 (Page 177) shows tasks ideally completed by computing devices.

Detailed Explanation

Tasks Suitable for Humans 1. Subjective Judgments Definition: Tasks that involve subjective judgment and interpretation are often better suited for humans. Example from Textbook: Determining the sentiment of a piece of text or assessing the credibility of a news source may require human judgment. Why humans are better: Humans can understand context Humans can interpret tone and nuance Humans can make judgments based on experience Humans can understand cultural differences Examples: Assessing the tone of an article (positive, negative, neutral) Determining if a source is credible

Evaluating the quality of a piece of writing Judging the appropriateness of content 2. Complex Decision-Making Definition: Tasks that require complex decision-making based on a deep understanding of context, ethics, and social norms are typically handled better by humans. Example from Textbook: Making decisions about content moderation on a social media platform often requires considering a range of cultural and ethical factors. Why humans are better: Humans understand ethics and morality Humans can consider cultural context Humans can weigh competing values Humans can make nuanced decisions Examples: Content moderation decisions Ethical decisions in healthcare Legal judgments Policy decisions 3. Handling Ambiguity Definition: Humans are capable of adapting and making decisions based on their knowledge and common sense when the data or task

requirements are unclear . Why humans are better: Humans can use common sense Humans can adapt to new situations Humans can make decisions with incomplete information Humans can ask clarifying questions Examples: Understanding unclear instructions Dealing with incomplete data Adapting to changing situations Making decisions with uncertainty 4. Personal Judgments Definition: Humans are frequently better equipped for tasks involving arbitrary judgment, interpretation, or nuance. Example from Textbook: Human judgment may be necessary to determine the tone of a text or judge the reliability of a news source. Why humans are better: Humans can make personal judgments Humans can interpret nuance Humans can understand subtlety Humans can apply personal experience

Examples: Judging the reliability of a news source Determining the tone of a text Evaluating the quality of art Assessing the appropriateness of a response Tasks Suitable for Computers 1. Speed and Reliability Definition: Computers can handle data far more quickly and with constant accuracy, which makes them the best choice for activities requiring frequent and repetitive calculations. Why computers are better: Computers are faster Computers are more reliable Computers do not get tired Computers do not make careless mistakes Examples: Calculating large datasets Processing millions of records Performing repetitive calculations Sorting and filtering data 2. Processing Large Amounts of Data Definition:

Handling large volumes of data quickly and accurately is a strength of computing devices. Why computers are better: Computers can process massive amounts of data Computers can handle data at scale Computers can perform tasks like sorting, filtering, and aggregating Computers can process data faster than humans Examples: Analyzing big data Processing millions of transactions Aggregating data from multiple sources Sorting large datasets 3. Statistical Analysis Definition: Computing devices are well-suited for performing complex statistical analyses, regression modeling, and other data-driven calculations. Why computers are better: Computers can perform complex calculations Computers can handle large datasets Computers can perform statistical tests Computers can create visualizations Examples: Regression analysis

Hypothesis testing ANOVA Correlation analysis 4. Pattern Recognition Definition: Tasks that involve pattern recognition, such as identifying anomalies in data or detecting trends, can be efficiently handled by machine learning algorithms. Why computers are better: Computers can find patterns in large data Computers can detect anomalies Computers can identify trends Computers can learn from data Examples: Fraud detection Anomaly detection Trend analysis Image recognition 5. Speed and Consistency Definition: Computing devices can process data much faster and with consistent precision, making them ideal for tasks that require rapid and repetitive calculations. Why computers are better:

Computers are fast Computers are consistent Computers do not get bored Computers do not make errors due to fatigue Examples: Real-time data processing Continuous monitoring Automated trading Quality control 6. Automated Verification Definition: Tasks like verifying data against predefined rules or performing data quality checks can be automated using computing devices. Why computers are better: Computers can check rules automatically Computers can verify data against standards Computers can perform quality checks Computers can do this at scale Examples: Data validation Quality assurance Compliance checking Format verification

Intermediate Level

In data science, both humans and computers are used together . Example: Content Moderation on Social Media Tasks for Computers: Scan millions of posts per day Flag posts with certain keywords Detect spam automatically Identify duplicate content Tasks for Humans: Review flagged posts Decide if content violates community guidelines Consider cultural context

Make final decisions on complex cases How They Work Together: Computers handle the volume Humans handle the nuance Together they moderate effectively

Important Points

• Data source verification tasks can be done by humans or computers.1.
• Choice depends on: Task nature, accuracy requirements, speed, resources.2.
• Tasks suitable for humans:3.
• Subjective judgments
• Complex decision-making
• Handling ambiguity
• Personal judgments
• Tasks suitable for computers:4.
• Speed and reliability
• Processing large amounts of data
• Statistical analysis
• Pattern recognition
• Speed and consistency
• Automated verification
• Figure 6.15 shows tasks suitable for humans.5.
• Figure 6.16 shows tasks ideally completed by computing devices.6.
• In many cases, a combination of both humans and computers is most effective.7.
• Humans provide expertise, judgment, and oversight.8.
• Computers handle repetitive and data-intensive tasks efficiently.9.

Remember

• Tasks for Humans:
• Subjective Judgments
• Complex Decision-Making
• Handling Ambiguity
• Personal Judgments
• Tasks for Computers:
• Speed and Reliability
• Processing Large Amounts of Data
• Statistical Analysis
• Pattern Recognition
• Speed and Consistency
• Automated Verification
• Figure 6.15 = tasks for humans
• Figure 6.16 = tasks for computers
• Combination of both is often most effective
• Humans: Expertise, judgment, oversight
• Computers: Repetitive, data-intensive tasks
• Warning: Not all tasks are suitable for computers

Common Mistakes

• Assuming computers can do everything – Some tasks need human judgment.1.
• Assuming humans are always better – Computers are better for speed and scale.2.
• Not considering task nature – Choose based on the task.3.
• Ignoring accuracy requirements – Some tasks need high4.