Social Networking Analysis

We the “tech-savvy” are dwelling in the world of Web 2.0 and are already on the verge of evolution into Web 3.0. The major factor that contributed to Web 2.0 is social networking – the influence of “people” connections on internet. Facebook is inevitably the major player, but not the first one. Sixdegrees.com launched in 1997 was the first to implement this concept. But this triggered off many online forums and chat rooms which became the sensation of the past decade. Any teenager not on facebook is literally considered to be “anti-social” or a “boring” person...!! It looks like a “really cool” stuff where I can find my second grade friend on facebook but there is a lot more than this. There are a lot of companies out there who try to analyze our networks and find some meaning out of it.

“Social network Analysis” is indeed not as fanciful or glamorous as you see. It involves a lot of calculations. It is the measuring of relationships between people or groups. A node in a network can represent a single person, a group, a tweet, hastag or a community. Edges are the connections between the nodes. The edges can be directed or undirected. For example in twitter you can know who is following whom, so it is a directed network. Apart from this there are “zillion” other terms that reveals different truths about the network. Following are a few question which would help you understand the terms.

Who is the hub for the network?

Degree centrality would give the answer for this question. It measures the most number of direct connections made by a particular node. The degree centrality of the network would be the node which has the maximum degree centrality. That node is the most active, that person is connected to every other person on the network.

Who can access all the other nodes more quickly?           

Closeness centrality of a node tells whether it can access all other nodes quickly or not. The closeness centrality of a network has the shortest paths to every other node in the network.  So the information can spread quickly through this node.

Who can act as a mediator in the network?

Betweeness Centrality gives the probability of a node occurring between any two nodes. So “he” is the person whom you can find between any two your friends.

Who is the most influential person in the network?

Eigen vector weighs every connection that a node has and the eigenvector centrality gives the weighted sum of all connections (direct as well as undirected). The person having the highest eigen vector is the most famous person in the network. 

Longtail keywords and definitions for Google ad words

Deciding on keywords is the most time consuming part of the Google adword marketing challenge. You need to think from the searchers perspective – a lot of mental work is involved. I’m pretty sure that “Long tail” keywords originated from the concept of long-tail – finding a market for obscure product. These keywords are usually 3 to 5 words long are rarely been used. But we need to flip the coin and look at the advantageous part of it. These keywords come for a lesser CPC. So you can more audience by having more number of long tail keywords. This is another marketing strategy..!

Example of some long tail keywords are: “ information security courses in Arizona”

There are a couple of terms that we would have to analyze in order to check the effectiveness of our adword. Following are a couple of them:

Avg CPC(Cost per Click): This is a simple average operation performed on the total no of clicks that your ad has received.

Total actual CPC / No of clicks

Actual CPC : The cost that you pay for each click on your ad and it can be lesser than  the maximum bid that you have set for the keyword. You will be charged only the amount will allow you to beat the adrank of your adword.

CTR – Click through Rate : This is a ratio of no of clicks to the number of impressions that your ad makes on the search result page. It is very important in deciding how well you have chosen the keywords and how effective is your bidding strategy

Impressions: Every time your ad shows up on the search result page it is counted as one impression.

Quality score: Every adword has a quality score which is determined by the potential of your keyword to trigger your adword.

Adrank: This is the value assigned to your adword based on quality score. Your adrank determines the ad position.

Wanna advertise online..Here you go..!

How does Google make money? If anyone of you have thought about it you would definitely know that Google makes its revenue from the online advertisement that we see popping on to the top and right(sometimes even at the bootom also) of the google search webpage. It seems very easy for us to look as a searcher to see the various ads being displayed on the search result page. But there is more than this that goes behind these ads. There are many different types of online ads – textual ads, display ads, pop-up ads, flash ads and image ads.

But in this blog we’ll be focussing only on textual ads. I would say that creation of a Google ad undergoes a complete life cycle from deciding on the theme to the bidding of these ads. Every company wants to attract the users by displaying the best deals/ offers on their product. So these ads are basically designed for any special event or deal. It is not necessary that a company should have just one adword, there can be many depending on the variety of products or deals. Once you have finalized on the theme of the adword you need to work on adword design. Every adword is made of 4 lines. As you can see from the image the first line is the header followed by the display url and the content of the ad. Be specific in the ad content because this is the only way to drive the user to your website. All that I have discussed above is the easiest of constructing an adword. The most challenging part is building a bidding strategy for the adword.

Every adword can be tagged with 40 to 50 keywords. These keywords are nothing but the possible textual content that any user will be typing in the google search box that is looking for a product that is related to our adword. So you need to put yourself in the searcher’s shoes and think of all possible combinations of keywords.  It doesn’t stop there, you still have the bidding strategy left for these keywords. There are two type of bidding strategies – Manual and Automatic. In manual bidding you will have to key in the bidding amount for each keyword based on the analysis that you would have done using Keyword tool and traffic estimator. These two tools come handy in deciding the bid amount because it shows the global monthly searches. In automatic bidding the bid values will be adjusted based on the number of clicks that your ad will be receiving and the ad position.

I think I have pretty much tried to explain the Google adwords on a high level in this blog.

Lecture 3 : Best/Better Testing

 Lecture 3:  A/B testing

Every software development cycle has got a testing phase. This is very important for any project to be successful because it is during this phase all unseen and undiscovered bugs of the software are detected. A/B testing is also very similar to other types of testing like unit, regression, functional – blackbox or whitebox testing. The name of this type of testing clearly tells any novice user that it means to pick from the two choices available A or B.

There can be various KPIs set for a website and there can be hundreds of metrics used to measure these KPIs.  A/B testing serves the same purpose of KPIs. It is used to determine the success of a website by comparing two versions of it along the same metrics. Whichever version sustains the race wins the battle. Let me further investigate into this topic : A and B are two versions or designs of a website. Usually A is the existing website and B is the newly designed one. Businesses do this to test the effectiveness of the new website. The traffic to the website is split between the two versions and the performance is measured based on the metrics you have set. Say you would like to have more people signing up for an event, you can measure it for the two versions and chose the best of it and promote that version of the website.

There are certain usual elements that are being tested using A/B testing: Headlines, Images, Price, Sign-up forms length. There are free tools available on the internet for performing this test. Google Website Optimizer is one such tool. You can use A/B testing to either test one feature (say just the Signup button) or the entire website.

I guess Google Website Optimizer tool will be handy and useful for our GOMC project.

  

Lecture 1: Buzziness Intelligence

And today was my first “lecture” but it did not seem to be, because of the overflowing posts and tweets that were flying since last Wednesday. As soon as I entered the class I could see a bunch of my classmates crowding near the TA table. I did not give it much of an importance because I was rather very much interested in securing a seat in the first or the second row (I usually tend to lose my concentration-(polished way of saying ”I fall asleep”) after the first half-an-hour lecture and BI is not a course where I’m supposed to do it...!). After settling down in the second row I heard Snehal calling me to ask whether I had signed the attendance sheet or not. Okaye so that long- queue was for the Attendance Sheet .(Attendance also counts for the 15%  for class participation).

Dr.Sudha Ram, my professor for MIS 587-Business Intelligence had arrived, three minutes before the clock ticked 8 o’clock. I could see her busily writing down the agenda for the class in a corner of the white board.  We settled down almost 5 minutes past 8 o’clock after everyone had signed the attendance sheet. And then my professor began her lecture with a simple question ( not really..!! ) “What is Business Intelligence?” I was trying to recollect what all I had read during the past 4 days from all the tweets and facebook posts and before I could frame my thoughts someone answered saying its “Collection of Unstructured Data”. Then I could hear people giving out random answers like “BI deals with historical data”, “BI is used for prediction”, ”BI helps identify trends”, ”BI helps gain competitive advantage for a company”..And finally the class came into a consensus after my professor summed up all the points and defined BI as “Collection,measurement,understanding,analysis and prediction using data for performance management.”

While talking about data she did stress on the differences between internal and external data. From what I understood is that Internal data is the data that is collected from the day to day operations of the business, something like the sales data of a particular product, transactional data in banks, enrolment data in school and so on. While external data is something that is gathered from forums, networking websites like facebook, twitter, blogs, etc. So for a business to predict its growth it definitely needs to analyse both these forms of data.

Someone did ask a very interesting question, “How do we rely on these data? What if people are just bluffing?” Pretty much an argumentative topic. After all its using the data from external sources we predict our business growth. So how do we trust these data? And then came the answer from our professor, that Data Profiling is the process of examining the data for its authenticity and purity.

{ My findings:  Data profiling is a method used to analyse data from data warehouses to clarify the structure, content, relationships and derivation rules of the data. The results of the analysis is used in making strategic business decisions. There are several different kinds of Data profiling techniques}

Data , Social Media and Prediction were the three constantly used words in class. I could hear it from every corner of my class. There was another intuitive question asked by one of my batch mates, “Why can’t we do all these collection, analysis and prediction using some JAVA algorithm?” Oh GOD are we going to talk about ALGOS in class now.!!  No..not now ..already there were a plenty of new terms piercing my cortex. Thankfully my professor had a simple and a straight answer to it “NO. Algorithms are a part of the techniques used in prediction or analysis of the data." 

“KPI  - Key Performance Indicator” was another important term that was under constantly probed during the class. KPIs are used for performance measurement.

And the time was 9 ‘o clock .Great !! I was pretty much awake and still listening to my class lecture. Then there were discussions on Paradigm shift and Web analytics which I’ll be discussing in my next blog. So stay tuned and frequently visit my blog until my next blog shows up.