What is Machine Learning? What is Artificial Intelligence? What is Big Data and Data Analysis?
People working in IT industry are bowled with these questions now a days.Are not computers intelligent enough to do almost all the tasks perceived by human beings? Why is it necessary that machine should learn? Are not our software engineers making them to learn to do any kind of task? Are computers capable of learning?
Massive amount of data is being stored and manipulated on the servers and in the data-centers. Data is being stored to convert it into information by various types of report generating algorithms. So what is Big Data?
Why further analysis of data is needed? What is Data Analysis?
I know these questions are lingering as doubts in your mind. In this article , there is an attempt to clear these doubts in a simple manner.
Computers are still the dumbest machines on the earth. That's why they require intelligent software engineers to Program them as per the task to be performed. They still do not have the capacity to think as human beings. If you write a computer program to join two alphanumeric text e.g. "A" and "B", the output will be "AB". When the program will be given input data as "2" and "4" , the output will be "24". The program does not have capacity to think that it can be a numeric addition (+) also. There is a possibility of answer (6). If you feed the above program with "CAR" and "TRUCK", the output will be "CARTRUCK". The program
does not have capacity to think and point that "CAR" and "TRUCK" can not be added in the real world.
All the above is based on the assumption that the program is a very basic program. It has been developed to add to alphanumeric text without having any further sophisticated problem solving logic built into it.
does not have capacity to think and point that "CAR" and "TRUCK" can not be added in the real world.
All the above is based on the assumption that the program is a very basic program. It has been developed to add to alphanumeric text without having any further sophisticated problem solving logic built into it.
Artificial Intelligence (AI) based programs attempt to solve this basic problem of not having intelligence.
Artificial Intelligence(AI) is a term given to a set of programs(based on AI algorithms) that builds intelligence
in the computer programs so that they evolve and learn when they are used with new set of data.So the responses of such programs vary time to time and they keep becoming more accurate as more real world data is given to them. These programs have capacity to modify their rules and problem solving logic for giving responses as new knowledge or data is given to them. Artificial Intelligence(AI) is a generic term given to such hardware or software or combination of both which demonstrates such capacity. AI based robot is the excellent example of such hardware and software.
in the computer programs so that they evolve and learn when they are used with new set of data.So the responses of such programs vary time to time and they keep becoming more accurate as more real world data is given to them. These programs have capacity to modify their rules and problem solving logic for giving responses as new knowledge or data is given to them. Artificial Intelligence(AI) is a generic term given to such hardware or software or combination of both which demonstrates such capacity. AI based robot is the excellent example of such hardware and software.
Machine Learning is also part of AI , but generally it applies to software components which work with big database of knowledge. It is a subset of AI. Machine Learning programs are capable of learning from the data presented to them by building realistic data models mimicking the real world.Their output becomes more realistic as big chunk of realistic data of real world problems is fed to them. Here the term Big Data comes into the picture. It is simply a term being used for vast amount of data for a particular domain or knowledge segment like finance, marketing, weather, geographical data,customer queries,etc.
What is a difference between a normal non-AI program and AI program?
A normal non-AI program can give predetermined answers to a predetermined set of questions for which it has been designed. Any question out of the set for which it is not designed will not be answered or taken.
An AI based program will treat questions out of set as new knowledge or data and it will modify itself and it's rules. It is called learning. Next time, if the same question is presented, it will give appropriate answer as per what it has learned. So AI programs are like little children. You keep giving them new information and they keep learning day by day. Their answers and responses changes day by day with new knowledge and insight in the direction of highest accuracy.
An AI based program will treat questions out of set as new knowledge or data and it will modify itself and it's rules. It is called learning. Next time, if the same question is presented, it will give appropriate answer as per what it has learned. So AI programs are like little children. You keep giving them new information and they keep learning day by day. Their answers and responses changes day by day with new knowledge and insight in the direction of highest accuracy.
Differences between non -AI and AI based programs or software
Normal non-AI based programs needs to be modified in order to solve new problems or to tackle new data or information. AI based programs have built-in capacity to modify themselves. Rarely any changes in programs are required. Only new knowledge or insight is required to modify it.
How E-commerce industry will be impacted?
E-commerce and E-tailing will greatly benefit with Machine Learning and AI. It is the industry where massive amount of data is generated day by day. On-line users or customers for E-commerce companies are highly intelligent and unpredictable in their buying patterns. They do not believe in loyalty. They are price-sensitive and prefer value for money. E-commerce portals do not give personal selling experience to on-line customers like brick and mortar stores. It is this basic lacuna where AI based stores will cater to.
Look and feel enhancement of store as per individual customer
AI based E-commerce stores will have intelligent look and feel or User Interface(UI) design. They will be adaptive in layout and theme color display. They will change layout and theme as per the taste of customer.
How will the E-commerce portal determine taste and preference of customer on individual basis? The answer lies into their product or service buying habit. If a customer consistently shops for apparels of red and blue colors, he/she is choosing simple designs of apparels with out any fancy items, the customer has preference for simplicity and taste for the above colors.
The layout will be adjusted from simple to complex as per their mouse clicking habit. If a customer is using simple widgets to perform the tasks on the store , it indicates that he/she is not tech savvy. If a customer is using complex features of the store, it indicates that he/she is tech savvy and AI based store will present advanced features of User Interface or UI to that customer.
All that is possible because AI algorithms will maintain distinct persona of each customer in their database. The distinct persona will consist of demographic information,buying habit,income level,family details,hobbies,passions,dreams and other relevant information. It will be codified as "IF THEN ELSE RULE" in the knowledge base of the store. It will be maintained per customer wise.
How will the E-commerce portal determine taste and preference of customer on individual basis? The answer lies into their product or service buying habit. If a customer consistently shops for apparels of red and blue colors, he/she is choosing simple designs of apparels with out any fancy items, the customer has preference for simplicity and taste for the above colors.
The layout will be adjusted from simple to complex as per their mouse clicking habit. If a customer is using simple widgets to perform the tasks on the store , it indicates that he/she is not tech savvy. If a customer is using complex features of the store, it indicates that he/she is tech savvy and AI based store will present advanced features of User Interface or UI to that customer.
All that is possible because AI algorithms will maintain distinct persona of each customer in their database. The distinct persona will consist of demographic information,buying habit,income level,family details,hobbies,passions,dreams and other relevant information. It will be codified as "IF THEN ELSE RULE" in the knowledge base of the store. It will be maintained per customer wise.
Product or service recommendations as per individual customer
Currently, product recommendations are defined by category manager or catalog manager of E-commerce
companies. It is not individualistic or personalized as per the single customer. So all customers find
same kind of recommendations delivered to them based on their selection of products in the cart.
companies. It is not individualistic or personalized as per the single customer. So all customers find
same kind of recommendations delivered to them based on their selection of products in the cart.
It will change drastically because of AI algorithms. Recommendations will be based on the taste of
individual customers and no two customers will be given same recommendations as their choices in the cart
may vary drastically. It is possible because AI algorithms will maintain distinct persona of each customer in their database. The distinct persona will consist of demographic information,buying habit,income level,family details,hobbies,passions,dreams and other relevant information.
individual customers and no two customers will be given same recommendations as their choices in the cart
may vary drastically. It is possible because AI algorithms will maintain distinct persona of each customer in their database. The distinct persona will consist of demographic information,buying habit,income level,family details,hobbies,passions,dreams and other relevant information.
Searching of products/services will be customized as per individual customer
Search results will be filtered as per customer's persona. Many other important factors which will be taken care by AI algorithms while producing search result. Following are the factors which algorithm may consider:
- Country from where customer is visiting the online store
- Date & Time of customer visit
- Any festivals are being celebrated in the country
- Any political situation or unrest is going in the country
AI algorithms will produce results based on all the above factors. Example: In India, during festival of light vacation, search will give weight to product/services related to celebrate Diwali festival, if customer is from that region. During Christmas vacation, search will give weight to products/services related to Christmas celebration. Again all the above suggestions will be further customized based on customer's demographic information,income criteria, passions, hobbies and other factors.
Anomaly Detection and Fraud Detection
E-commerce companies generate massive amount of data due to their day to day operations. Following types of data are generated:
- New customer profiles
- Order detail of new and old customers
- Banking Details of new and old customers, credit card, debit card and Net Banking details except sensitive information
- Addresses and delivery information where geo-graphic data is also part of it.
E-commerce companies have to deal frequently with abusive or fraud customers who make excessive or
fraud orders with stolen credit and debit cards or banking details. It is a major source of financial loss and tarnishing of reputation for these companies.
AI algorithms can maintain a common database for fraud detection collectively pooled with member E-commerce companies or E-tailing companies. Just like in INDIA , CIBIL maintains a list of bank defaulters which is a co-operative umbrella organization for all the banks and financial companies in INDIA.
The above database can use to detect fraud customers or users visiting the website of online store.
fraud orders with stolen credit and debit cards or banking details. It is a major source of financial loss and tarnishing of reputation for these companies.
AI algorithms can maintain a common database for fraud detection collectively pooled with member E-commerce companies or E-tailing companies. Just like in INDIA , CIBIL maintains a list of bank defaulters which is a co-operative umbrella organization for all the banks and financial companies in INDIA.
The above database can use to detect fraud customers or users visiting the website of online store.
Application in Customer Support
On-line customers are made to wait long for resolving queries because of limitation of human resources in customer support or call centers. A significant amount of Operating Expenditure (OPEX) is required to provide desirable lever of customer support and service.
AI algorithms and Machine Learning algorithms are going to reduce that. Intelligent chat bots or software robots will answer phone calls and they will solve the basic problems or queries of the customers. To solve more complex queries or problems, they will escalate the issue to human customer support.
Chat bots and software robots will keep enhancing their knowledge of customer problems day by day as they will interact with more and more customers. Speech recognition and natural language processing via. deep learning will play a major role into this.
Apart from phone calls, intelligent chat bots and software robots will automatically answer phone calls,categorize emails, complain, request for information. They will also provide feedback content to various other departments of the organization.
AI algorithms and Machine Learning algorithms are going to reduce that. Intelligent chat bots or software robots will answer phone calls and they will solve the basic problems or queries of the customers. To solve more complex queries or problems, they will escalate the issue to human customer support.
Chat bots and software robots will keep enhancing their knowledge of customer problems day by day as they will interact with more and more customers. Speech recognition and natural language processing via. deep learning will play a major role into this.
Apart from phone calls, intelligent chat bots and software robots will automatically answer phone calls,categorize emails, complain, request for information. They will also provide feedback content to various other departments of the organization.
Adaptive Pricing or Dynamic pricing of products/services
The biggest achievement of AI algorithms for E-commerce or E-tailing companies will be sensing the customer's pocket or capacity to pay. Algorithms will sense the market conditions for a particular product or service based on number of factors like prices offered by competitive companies,season, demand-supply gap for a particular product or service, sales geography and region and future demand. AI algorithms will suggest prices to E-commerce vendors based on the above factors. Also , they will help in maintaining optimum inventory based on market dynamics and demand-supply gap.
Summary
AI is a promising technology for E-commerce or E-tailing vendors to create WOW factor in their on-line stores. Let us keep our fingers crossed.
