Assignment
Discussion 1 (Chapter 12): Examine Alexas skill in ordering drinks from Starbucks.
Discussion 2 (Chapter 13): Research Apple Home Pod. How does it interact with smart home devices? Alexa is now connected to smart home devices such as thermostats and microwaves. Find examples of other appliances that are connected to Alexa and write a report.
Each discussion should be 250-300 words.
There must be at least one APA formatted reference (and APA in-text citation) to support the thoughts in the post. Do not use direct quotes, rather rephrase the author’s words and continue to use in-text citations.
Chapter12Knowledge Systems: Expert Systems, Recommenders, Chatbots, Virtual Personal Assistants, and Robo Advisors
Learning Objectives
Describe recommendation systems
Describe expert systems
Describe chatbots
Understand the drivers and capabilities of chatbots and their use
Describe virtual personal assistants and their benefits
Describe the use of chatbots as advisors
Discuss the major issues related to the implementation of chatbots
Advancement in artificial intelligence (AI) technologies and especially natural language processing (NLP), machine and deep learning and knowledge systems, coupled with the increased quality and functionalities of other intelligent systems, and mobile devices and their apps, have driven the development of chatbots (bots) for inexpensive and fast execution of many tasks related to communication, collaboration, and information retrieval. The use of chatbots in business is increasing rapidly, partly because of their fit with mobile systems and devices. As a matter of fact, sending messages is probably the major activity in the mobile world.
In the last two to three years, many thousands of bots have been placed into service worldwide by both organizations (private and public) and individuals. Many people refer to these phenomena as thechatbot revolution. Chatbots today are much more sophisticated than those of the past. They are extensively used, for example, in marketing; customer, government, and financial services; healthcare; and in manufacturing. Chatbots make communication more personal than faceless computers and excel in data gathering. Chatbots can stand alone or be parts of other knowledge systems.
We divide the applications in this chapter into four categories: expert systems, chatbots for communication and collaboration, virtual personal assistants (native products, such as Alexa), and chatbots that are used as professional advisors. Some implementation topics of intelligent systems are described last.
This chapter has the following sections:
1. 12.1Opening Vignette: Sephora Excels with Chatbots649
2. 12.2Expert Systems and Recommenders650
3. 12.3Concepts, Drivers, and Benefits of Chatbots660
4. 12.4Enterprise Chatbots664
5. 12.5Virtual Personal Assistants672
6. 12.6Chatbots as Professional Advisors (Robo Advisors)676
7. 12.7Implementation Issues680
12.1Opening Vignette:Sephora Excels with Chatbots
The problem
Sephora is a French-based cosmetics/beauty products company doing business globally. It has its own stores and sells its goods in cosmetic and department stores. In addition, Sephora sells online on Amazon and on its online store. The company sells hundreds of brands, including many of its own. It operates in a very competitive market where customer care and advertising are critical. Sephora sells some products for men, but most beauty products are targeted to women.
The Solution
Sephoras first use of chatbots occurred through messaging services. The purpose of the first bot was to search for information for the companys resources such as videos, images, tips, and so on. This bot operates in a question-and-answer (Q&A) mode. It recommends relevant content based on customers interests. The company aims to appeal to young customers messaging on Kik.
Sephora researchers found that customers conversing with the Kikbot were engaged deeply in the dialog. Then the bot encouraged them to explore new products. Sephoras newer bot called Reservation Assistant was placed on Facebook Messenger. It enables customers to book or reschedule makeover appointments.
Another Sephora bot delivered on Kik is Shade-Matching. It matches lips colors to photos (face and lips) uploaded by users and recommends the best match to them. The bot also lets users try on photos of recommended colors, using Sephora Virtual Artist that runs on Facebook Messenger. Bots are deployed as mobile apps. If users like the recommendation, they are directed to the companys Web store to buy the products. Users can upload photos taken with selfies so that the program can do the matching. Over 4 million visitors tried 90 million shades in the first year of Virtual Artists operation.
The Q&A collection of the knowledge base was built by connecting it with store experts. Knowledge acquisition techniques (
Chapter2
) were used for this purpose. The companys bots use NLPs that were trained to understand the typical vocabulary of users.
The Results
The companys customers loved the bots. In addition, Sephora learned the importance of providing assistance and guidance to users who are motivated to return (at a reasonable cost!), happier, and more engaged.
Sephoras bot asks users questions to find their tastes and preferences. Then it acts like arecommendation system(
Section12.2
), offering products. Kik and Messenger users can purchase items without leaving the messaging service.
Finally, the company has improved the bots knowledge over time and plans new bots for additional tasks.
NOTE:
Sephora was selected byFast Company Magazine, March/April 2018, as one of the Worlds Most Innovative Companies. Sephora is known for its digital transformation and innovation (Rayome, 2018). Also, Sephoras bots are considered among the top marketing chatbots (
Quoc, 2017
).
Sources:
Compiled from
Arthur (2016)
,Rayome (2018), and
Taylor (2016)
,
theverge.com/2017/3/16/14946086/sephora-virtual-assistant-ios-app-update-ar-makeup/
, and
sephora.com/
.
Questions for the Opening Vignette
1. List and discuss the benefits of bots to the company.
2. List and discuss the benefits of bots to customers.
3. Why were the bots deployed via Messenger and Kik?
4. What would happen to Sephora if competitors use a similar approach?
What We Can Learn from This Vignette
In the highly competitive world of retail beauty products, customer care and marketing are critical. Using only live employees can be very expensive. In addition, customers are shopping24/7,24/7,and physical stores are open during limited hours and days. In addition, there are large combinations of certain beauty products (e.g., many shades/colors) available. Sephora decided to use chatbots on Facebook Messenger and Kik to engage its customers. Chatbots, the subject of this chapter, are available24/724/7at a lower cost and are delivered via mobile devices. Bots deliver information to customers consistently and quickly direct customers to easy online shopping. Sephora placed its chatbots on messaging services. The logic was that people like to chat with friends on messaging services, and they may also like to chat with businesses.
In addition to several services to customers, using chatbots helps Sephora learn about customers. This type of chatbot is the most common type for customer care and marketing. In this chapter, we cover several other types of knowledge systems, including the pioneering expert systems, recommenders, virtual personal assistants offered by several large technology companies, and robo advisors.
12.2Expert Systems and Recommenders
InChapter2we introduced the reader to the concept of autonomous decision systems. Anexpert systemis a category of autonomous decision systems and are considered the earliest applications of AI. Expert systems use started in research institutions in the early and mid-1960s (e.g., Stanford University, IBM) and was adopted commercially during the 1980s.
Basic Concepts of Expert Systems (ES)
The following are the major concepts related to ES technology.
Definitions
There are several definitions of expert systems. Our working definition is that an
expert system
is a computer-based system that emulates decision making and/or problem solving of human experts. These decisions and problems are in complex areasthat require expertise to solve. The basic objective is to enable nonexperts to make decisions and solve problems that usually require expertise. This activity is usually performed in narrowly defined domains (e.g., making small loans, providing tax advice, analyzing reasons for machine failure). Classical ES use what-if-then rules for their reasoning.
Experts
An
expert
is a person who has the special knowledge, judgment, experience, and skills to provide sound advice and solve complex problems in a narrowly defined area. It is an experts job to provide the knowledge about how to perform a task so that a nonexpert will be able to do the same task assisted by ES. An expert knows which facts are important and understands and explains the dependent relationships among those facts. In diagnosing a problem with an automobiles electrical system, for example, an expert car mechanic knows that a broken fan belt can be the cause for the battery to discharge.
There is no standard definition ofexpert, but decision performance and the level of knowledge a person has are typical criteria used to determine whether a particular person is an expert as related to ES. Typically, experts must be able to solve a problem and achieve a performance level that is significantly better than average. An expert at one time or in one region may not be an expert in another time or region. For example, a legal expert in New York may not be one in Beijing, China. A medical student may be an expert compared to the general public but not in making a diagnosis or performing surgery. Note that experts have expertise that can help solve problems and explain certain obscure phenomena only within a specific domain.
Typically, human experts are capable of doing the following:
Recognizing and formulating a problem.
Solving a problem quickly and correctly.
Explaining a solution.
Learning from experience.
Restructuring knowledge.
Breaking rules (i.e., going outside the general norms) if necessary.
Determining relevance and associations.
Can a machine help a nonexpert perform like an expert? Can a machine make autonomous decisions that experts make? Let us see. But first, we need to explore what expertise is.
Expertise
An
expertise
is the extensive, task-specific knowledge that experts possess. The level of expertise determines the success of a decision made by an expert. Expertise is often acquired through training, learning, and experience in practice. It includes explicit knowledge, such as theories learned from a textbook or a classroom and implicit knowledge gained from experience. The following is a list of possible knowledge types used in ES applications:
Theories about the problem domain.
Rules and procedures regarding the general problem domain.
Heuristics about what to do in a given problem situation.
Global strategies for solving of problems amenable to expert systems.
Meta knowledge (i.e., knowledge about knowledge).
Facts about the problem area.
These types of knowledge enable experts to make better and faster decisions than nonexperts.
Expertise often includes the following characteristics:
It is usually associated with a high degree of intelligence, but it is not always associated with the smartest person.
It is usually associated with a vast quantity of knowledge.
It is based on learning from past successes and mistakes.
It is based on knowledge that is well stored, organized, and quickly retrievable from an expert who has excellent recall of patterns from previous experiences.
Characteristics and Benefits of ES
ES were used during the period 1980 to 2010 by hundreds of companies worldwide. However, since 2011, their use has declined rapidly, mostly due to the emergence of better knowledge systems, three types of which are described in this chapter. It is important, however, to understand the major characteristics and benefits of expert systems since many of them evolved evidenced newer knowledge systems.
The major objective of ES is the transfer of expertise to a machine. The expertise will be used by nonexperts. A typical example is a diagnosis. For example, many of us can use self-diagnosis to find (and correct) problems in our computers. Even more than that, computers can find and correct problems by themselves. One field in which such ability is practiced is medicine, as described in the following example:
Example: Are You Crazy?
A Web-based ES was developed in Korea for people to self-check their mental health status. Anyone in the world can access it and get a free evaluation. The knowledge for the system was collected from a survey of 3,235 Korean immigrants. The results of the survey were analyzed and then reviewed by experts via focus group discussions. For more information, seeBae (2013).
Benefits of ES
Depending on the mission and structure of ES, the following are their capabilities and potential benefits:
Perform routine tasks (e.g., diagnosis, candidate screening, credit analysis) that require expertise much faster than humans.
Reduce the cost of operations.
Improve consistency and quality of work (e.g., reduce human errors).
Speed up decision making and make consistent decisions.
May motivate employees to increase productivity.
Preserve scarce expertise of retiring employees.
Help transfer and reuse knowledge.
Reduce employee training cost by using self-training.
Solve complex problems without experts and solve them faster.
See things that even experts sometimes miss.
Combine expertise of several experts.
Centralize decision making (e.g., by using the cloud).
Facilitate knowledge sharing.
These benefits can provide a significant competitive advantage to companies that use ES. Indeed, some companies have saved considerable amounts of money using them.
Despite these benefits, the use of ES is on the decline. The reasons for this and the related limitations are discussed later in this section.
Typical Areas for ES Applications
ES have been applied commercially in a number of areas, including the following:
FINANCE.Finance ES include analysis of investments, credit, and financial reports; evaluation of insurance and performance; tax planning; fraud prevention; and financial planning.
DATA PROCESSING.Data processing ES include system planning, equipment selection, equipment maintenance, vendor evaluation, and network management.
MARKETING.Marketing ES include customer relationship management, market research and analysis, product planning, and market planning. Also, presale advice is provided for prospects.
HUMAN RESOURCES.Examples of human resource ES are planning, performance evaluation, staff scheduling, pension management, regulatory advising, and design of questionnaires.
MANUFACTURING.Manufacturing ES include production planning, complex product configuration, quality management, product design, plant site selection, and equipment maintenance and repair (including diagnosis).
HOMELAND SECURITY.These ES include terrorist threat assessment and terrorist finance detection.
BUSINESS PROCESS AUTOMATION.ES have been developed for desk automation, call center management, and regulation enforcement.
HEALTHCARE MANAGEMENT.ES have been developed for bioinformatics and other healthcare management issues.
REGULATORY AND COMPLIANCE REQUIREMENTS.Regulations can be complex. ES are using a stepwise process to ensure compliance.
WEB SITE DESIGN.A good Web site design requires paying attention to many variables and ensures that performance is up to standard. ES can lead to a proper design process.
Now that you are familiar with the basic concepts of ES, it is time to look at the internal structure of ES and how their goals are achieved.
Structure and Process of ES
As you may recall fromSection2.5andFigure2.5, the process of knowledge extraction and its use is divided into two distinct parts. In ES we refer to these as thedevelopment environmentand theconsultation environment(seeFigure12.1). An ES builder builds the necessary ES components and loads the knowledge base with appropriate representation of expert knowledge in thedevelopment environment. A nonexpert uses theconsultation environmentto obtain advice and solve problems using the expert knowledge embedded into the system. These two environments are usually separated.
Major Components of ES
The major components in typical expert systems include:
KNOWLEDGE ACQUISITION.Mostly from human experts, is usually obtained by knowledge engineers. This knowledge, which may derive from several sources, is integrated, validated, and verified.
KNOWLEDGE BASE.This is a knowledge repository. The knowledge is divided into knowledge about the domain and knowledge about problem solving and solution procedures. Also, the input data provided by the users may be stored in the knowledge base.
KNOWLEDGE REPRESENTATION.This is frequently organized as business rules (also known asproduction rules).
INFERENCE ENGINE.Also known as thecontrol structureor therule interpreter, this is the brain of ES. It provides the reasoning capability, namely the ability to answer users questions, provide recommendations for solutions, generate predictions, and conduct other relevant tasks. The engine manipulates the rules by either forward chaining or backward chaining. In 1990s ES started to use other inference methods.
USER INTERFACE.This component allows user inference engine interactions. In classical ES, this was done in writing or by using menus. In todays knowledge systems, it is done by natural languages and voice.
These major components of ES generate useful solutions in many areas. Remember that these areas need to be well structured and in fairly narrow domains. Less common is ajustifier/explanationsubsystem that shows users of rule-based systems the chains of rules used to arrive at conclusions. Also, least common is aknowledge refining subsystemthat helped to improve knowledge (e.g., rules) when new knowledge is added.
A major provider of expert systems technologies was Exsys Inc. While the company is no longer active in this business, its Web site (Exsys.com) is. It contains tutorials and a large number of cases related to its major software product,Exsys Corvid.Application Case12.1is one example.
Application Case12.1ES Aid in Identification of Chemical, Biological, and Radiological Agents
Terrorist attacks using chemical, biological, or radiological (CBR) agents are of great concern due to their potential for leading to large loss of life. The United States and other nations have spent billions of dollars on plans and protocols to defend against acts of terrorism that could involve CBR. However, CBR covers a wide range of input agents with many specific organisms that could be used in multiple ways. Timely response to such attacks requires rapid identification of the input agents involved. This can be a difficult process involving different methods and instruments.
The U.S. Environmental Protection Agency (EPA) along with Dr. Lawrence H. Keith, president of Instant Reference Sources Inc. and other consultants, have incorporated their knowledge, experience, and expertise as well as information in publicly available EPA documents to develop the CBR Advisor using Exsys Inc.s Corvid software.
One of the most important parts of the CBR Advisor is providing advice in logical step-by-step procedures to determine the identity of a toxic agent when little or no information is available, which is typical at the beginning of a terrorist attack. The system helps response staff proceed according to a well-established action plan even in such a highly stressful environment. The systems dual screens present three levels of information: (1) a top/executive level with brief answers, (2) an educational level with in-depth information, and (3) a research level with links to other documents, slide shows, forms, and Internet sites. CBR Advisors content includes:
How to classify threat warnings.
How to conduct an initial threat evaluation.
What immediate response actions to take.
How to perform site characterization.
How to evaluate the initial site and safe entry to it.
Where and how to best collect samples.
How to package and ship samples for analysis.
Restricted content includes CBR agents and methods for analyzing them. The CBR Advisor can be used for incident response and/or training. It has two different menus, one for emergency response and another, longer menu for training. It is a restricted software program and is not publicly available.
Questions for Case12.1
1. How can the CBR Advisor assist in making quick decisions?
2. What characteristics of the CBR Advisor make it an expert system?
3. What could be other situations in which similar expert systems can be employed?
Expert systems are also used in high-pressure situations in which human decision makers often need to take split-second actions involving both subjective as well as objective knowledge in responding to emergency situations.
Sources:www.exsys.comIdentification of Chemical, Biological and Radiological Agentshttp://www.exsyssoftware.com/CaseStudySelector/casestudies.html. April 2018. (Publicly available information.) Used with permission.
Why the Classical Type of ES Is Disappearing
The large benefits described earlier drove the implementation of many ES worldwide. However, like many other technologies, the classical ES have been replaced by better systems. Let us first look at some of the limitations of ES that contributed to its declining use.
1. The acquisition of knowledge from human experts has proven to be very expensive due to the shortage of good knowledge engineers as well as the possible need to interview several experts for one application.
2. Any acquired knowledge needed to be updated frequently at a high cost.
1. The rule-based foundation was frequently not robust and not too reliable or flexible and could have too many exceptions to the rules. Improved knowledge systems usedata-driven and statistical approaches to make the inferences with better success. In addition, case-based reasoning could work better only if a sufficient number of similar cases were available. So, usually it cannot support ES.
2. The rule-based user-interface needed to be supplemented (e.g., by voice communication, image maps). This could make ES too cumbersome.
3. The reasoning capability of rule-based technology is limited compared to use of newer mechanisms such as those used in machine learning.
New Generation of Expert SYSTEMS
Instead of using the old knowledge acquisition and representation system, newer ES based on machine learning algorithms and other AI technologies are deployed to create better systems. An example is provided inApplication Case12.2.
Application Case12.2VisiRule
VisiRule is an older ES company that remodeled its business over time. VisiRule (of the United Kingdom) provides easy-to-use diagramming tools to facilitate the construction of ES. Diagramming allows easier extraction and use of knowledge in expert systems.
The process of building the knowledge base can be seen on the left side ofFigure12.2. On the left-hand side, you can see the hybrid creation. Using a decision tree, the domain experts can create additional rules directly from relevant data (e.g., historical). In addition, rules can be created by machine learning (lower left side).
n). Using interactive questions and answers the system can generate advice. In addition, rules can be used to process data remotely and update the data repository. Note that the dual delivery option is based on machine learnings ability to discover hidden patterns in data that can be used to form predictive decision models.
VisiRule also provides chatbots for improving the interactive part of the process and supplies an interactive map. According to the companys Web sitevisirule.co.uk/, the major benefits of the product are:
It is code-free; no programming is needed.
The diagrams are drawn by human experts or induced automatically from data.
It contains self-assessment tools with report generation and document production.
The generated knowledge can be easily executed as XML code.
It provides explanation and justification.
The interactive expert advice attracts new customers.
It can be used for training and advising employees.
Companies can easily access the corporate knowledge repository.
The charts to use VisiRule authoring tools are created with ease using flowcharting and decision trees.
The charts allow creation of models that can be immediately executed and validated.
All-in-all, VisiRule provides a comprehensive AI-based expert system.
Source:Courtesy of VisiRule Corp. UK. Used with permission.
Questions for Case12.2
1. Which of the limitations of early ES have been solved by the VisiRule system?
2. CompareFigures12.2and12.1. What are the differences between the creation (Fig.12.2) and the development (Fig.12.1) subsystems?
3. CompareFigures12.2and12.1. What are the differences between the delivery (Fig.12.2) and the consultation (Fig.12.1) subsystems?
4. Identify all AI technologies and list their contribution to the VisiRule system.
5. List some benefits of this ES to users.
Three major AI types of applications that overcome the earlier discussed limitations of RS are chatbots, virtual personal assistants, and robo advisors, which are presented next in this chapter. Other AI technologies that perform similar activities are presented inChapters4to 9. Most notable is IBM Watson (Chapter6); some of its advising capabilities are similar to those of ES but are much superior.
Another similar AI technology, the recommendation system, is presented next. Its newer variations use machine learning and IBM Watson Analytics.
Recommendation Systems
A heavily used knowledge system for recommending one-to-one targeted products or services is the
recommendation system
, also known asrecommender systemorrecommendation engine. Such a system tries to predict the importance (rating or preference) that a user will attach to a product or service. Once the rating is known, a vendor knows users tastes and preferences and can match and recommend a product or service to the user. For comprehensive coverage, seeAggarwal (2016). For a comprehensive tutorial and case study, seeanalyticsvidhya.com/blood/2015/10/recommendation-engines/.
Recommendation systems are very common and are used in many areas. Top applications includemovies, music, andbooks. However, there are also systems fortravel, restaurants, insurance, andonline dating. The recommendations are typically given in rank order. Online recommendations are preferred by many people over regular searches, which are less personalized, slower, and sometimes less accurate.
Benefits of Recommendation Systems
Using these systems may result in substantial benefits both to buyers and sellers (seeMakadia, 2018).
Benefits to customers are:
PERSONALIZATION.They receive recommendations that are very close to fulfilling what they like or need. This depends, of course, on the quality of the method used.
DISCOVERY.They may receive recommendations for products that they did not even know existed but were what they really need.
CUSTOMER SATISFACTION.With repeated recommendations tends to increase.
REPORTS.Some recommenders provide reports and others provide explanations about the selected products.
INCREASED DIALOG WITH SELLERS.Because recommendations may come with explanations, buyers may want more interactions with the sellers.
Benefits to sellers are:
HIGHER CONVERSION RATE.With personalized product recommendations, buyers tend to buy more.
INCREASED CROSS-SELL.Recommendation systems can suggest additional products.Amazon.com, for example, shows other products that people bought together with the product you ordered.
INCREASED CUSTOMER LOYALTY.As benefits to customers increase, their loyalty to the seller increases.
ENABLING OF MASS CUSTOMIZATION.This provides more information on potential customized orders.
Several methods are (or were) used for building recommendation systems. Two classic methods arecollaborative filteringandcontent-based filtering.
Collaborative Filtering
This method builds a model that summarizes the past behavior of shoppers, how they surf the Internet, what they were looking for, what they have purchased, and how much they like (rate) the products. Furthermore, collaborative filtering considers what shoppers with similar profiles bought and how they rated their purchases. From this, the method uses AI algorithms topredictthe preference of both old and new customers. Then, the computer program makes a recommendation.
Content-Based Filtering
This technique allows vendors to identify preferences by the attributes of the product(s) that customers have bought or intend to buy. Knowing these preferences, the vendor recommends to customers products with similar attributes. For instance, the system may recommend a text-mining book to a customer who has shown interest in data mining, or action movies after a consumer has rented one in this category.
Each of these types has advantages and limitations (see example aten.wikipedia.org/wiki/Recommender_system). Sometimes the two are combined into a unified method.
Several other filtering methods exist. Examples include rule-based filtering and activity-based filtering. Newer methods include machine learning and other AI technologies, as illustrated inApplication Case12.3.
Application Case12.3Netflix Recommender: A Critical Success Factor
According toir.netflix.com, Netflix is (Spring 2018 data) the worlds leading Internet television network with more than 118 million members in over 190 countries enjoying more than 150 million hours of TV shows and movies per day, including original series, documentaries, and feature films. Members can view unlimited shows without commercials for a monthly fee.
The Challenges
Netflix has several million titles and now produces its own shows. The large titles inventory often creates a problem for customers who have difficulty determining which offerings they want to watch. An additional challenge is that Netflix expanded its business from the United States and Canada to 190 other countries. Netflix operates in a very competitive environment in which large players such as Apple,Amazon.com, and Google operate. Netflix was looking for a way to distinguish itself from the competition by making useful recommendations to its customers.
The Original Recommendation Engine
Netflix originally was solely a mail-order business for DVDs. At that time, it encountered inventory problems due to its customers difficulties in determining which DVDs to rent. The solution was to develop a recommendation engine (called Cinematch) that told subscribers which titles they probably would like. Cinematch used data mining tools to sift through a database of billions of film ratings and customers rental histories. Using proprietary algorithms, it recommended rentals to customers. The recommendation was accomplished by comparing an individuals likes, dislikes, and preferences against those of people with similar tastes, using a variant ofcollaborative filtering. Cinematch was like the geeky clerk at a small movie store who sets aside titles he knows you will like and suggests them to you when you visit the store.
To improve Cinematchs accuracy, Netflix began a contest in October 2016, offering$1$1million to the first person or team that will write a program that would increase Cinematchs prediction accuracy by at least 10 percent. The company understood that this would take quite some time; therefore, it offered a$50,000$50,000Progress Prize each year in which the contest was conducted. After more than two years of competition, the grand prize went to Bellkors Pragmatic Chaos team, a combination of two runner-up teams.
To learn how the movie recommendation algorithms work, seequora.com/How-does-the-Netflix-movie-recommendation-algorithm-work/.
The New Era
As time passed, Netflix moved to the streaming business and then to Internet TV. Also, the spread ofcloud technologyenabled improvement in the recommendation system. The new system stopped making recommendations based on whatpeople have seen in the past. Instead, it is using Amazons cloud to mimic the human brain in order to find what people really like in their favorite movies and sho
