Chapter 5 Flashcards
What are the two types of filtering?
- Content-Based
- Collaborative
Content-Based Filtering: Idea
recommends items based on their features and the user’s preferences for those features
Content-Based Filtering: on what does it rely?
on a profile of the user’s prefferneces
Content-Based Filtering: Process
- build a profile for each user based on their preferences for different features of items
- recommend items that match the user’s profile and have features similar to what they have liked in the past
Content-Based Filtering: Pros
- addresses the cold start problem for new items
- can provide explanations fo recommendations based on item features
Content-Based Filtering: Cons
- may not capture complex user preferences that go beyond item features
- requires detailed information about the items and user preferences for features
Collaborative Filtering: Idea
makes recommendations based on user behavior and preferences
What does Collaborative Filtering assume?
that users who have agreed in the past tend to agree in the future
What are the types of Collaborative Filtering?
- User-Based
- Item-Based
How does User-Based Collabroative Filtering recommend items?
based on the preferences of users who are similar to the target user
How does Item-Based Collaborative Filtering recommend items?
it recommends items that are similar to those liked by the target user
Collaborative Filtering: Pros
- does not require knowledge of item features
- can capture complex patterns and user preferences
Collaborative Filtering: Cons
- cold start problem
- sparsity
Define the cold start problem
it can be challenging for Collaborative Filtering to provide accurate recommendations for new users or items with little or no history
Define Sparsity
Collaborative Filtering: when dealing with a large number of users and items, the user-item interaction matrix can be sparse, making it difficult to find similar users or items
Define User based collaborative filtering
similar tastes in the past
will have similar tastes in the future
Formula pred(Alice,Item5)
Formula sim(Item5, Item4)
Commonly used techniques of content-based filtering
- TF-IDF
- Clustering
- Decision trees
- ANN
- Bayesian networks
Well known problems of content-based filtering
- Cold-start problem for new users
- Cold-start problem for new items
- Scalability
- Sparsity problem
- Exotic profiles
Chatbot definition
computerized service that enables easy conversations between humans and humanlike computerized robots
Process of chatting with bots (graph)
Chatbots: categories
- enterprise chatbots
- virtual personal assistants
- Alexda
- Apple Siri
- Cortana - Robo-advisors
- financial advisor
- shopping advisor
- travel
- medical advisor
Chatbot: IBM estimates
265 billion service requests every year, addressing them costs companies 1.3 trillion dollars
What percentage of customer interactions wil be AI-powered by 2025?
95%
What percentage of customer service encounters in 2020 do not involve humans on the firm side?
85%
What is a chatbot building platform?
a chatbot development platform that does not require coding skills of developers
Definition of DSS by Gorry and Scott-Morton
interactive computer based systems which support decision makers by utilization of data and models to solve semistructured problems
List components of Decision Support Systems
- interactive computer-based systems
- decision makers
- decision support
- data
- models
- decision problems
What are the three staged of DSS as Judge-Advisor Systems?
- Input
- Process
- Output
Of what does the Input (stage of DSS as Judge-Advisor Systems) consist?
- advisor characteristics
- judge characteristics
- environment characteristics
Of what does the Process stage of DSS as Judge-Advisor Systems) consist?
- format of advice
- type of interaction
- explanation of advice
Of what does the Output stage of DSS as Judge-Advisor Systems) consist?
- trust
- accuracy
- confidence
- advice utilization
- system satisfaction
- intention to continue (use)
What is a Judge?
decision maker
What could an Advisor be?
- human
-alogirthm - human-algorithm team
What is Advice?
recommendation favoring or discouraging particular option(s)
What does Algorithm Aversion mean?
reluctance of human decision makers to use superior but imperfect algorithms
Why do people often chose human forecasters voer statistical algorithms?
- they lose confidence in algorithmic forecasters after seeing them make the same mistakes
- they would rather choose human even if the algorithm outperforms
What is algorithm appreciation?
adherence to advice when it comes from an algorithm than from a person
Measurement of algorithmic literacy (chart)
What is the solution to the problem: lack of decision control
human-in-the-loop decision making
What is the solution to the problem: lack of incentivization
behavioral design
What is the solution to the problem: combating intuition
engaging intuition
Describe algorithmic bias, fainress and transparency
- ai based predicions are faster, cheaper, more reliable and scalable
- unintended effects: discrimination or racism
What are sources for bias?
- biased training sets
- algorithm itself
- presentation formats
- users
What are legal, privacy and ethical issues?
- everyone may be affected by these applciations
- doable does not equal appropriate, legal or ethical
- data science and AI professionals/,amager must be aware of these concerns
What are corrective actions (for legal problems)?
- use unbiased data
- mandatory data governance
- model evaluation by social groups
- explainable AI/machine learning: from black box to glass box
Dark side of analytics; legal question
Who is liable for wrong advice?
Dark side of analytics: legal issues
- who owns the knowledge in a knowledge base?
- can management force experts to contribute to their expertise to an intelligent system?
Define Privacy
the right to be left alone and the right to be free from unreasonable personal intrusions
Is the right of privacy Absolut?
No
Which one is superior: the public’s right to know or the individual’s right to privacy?
the public’s right to know
What is AI’s impact on jobs?
- cuts jobs
- cuts opportunity
How can one deal with change
- use learning and education to facilitate the change
- involve the private sector in enhancing retraining
- have governments provide incentives to the private sector to improve human capital
- encourage private and public sectors to create appropiate digital infrastructure
- develop innovative income and wage schemes
- carefully plan the transition to the new work
- deal properly with displaced employees
