Dots Connector (DC) is a set of tools designed to analyze social networks created within social network services (such as Facebook, Nasza-klasa.pl or Grono.net), but it can also handle data from other (including off-line) sources.

DC treats networks of connections between people as its main source of information, it connects, splits, compares the data to show how much we say about ourselves just by adding friends to our buddies lists.

The main purpose of developing Dots Connector is sociological research, but it can also be educational and entertaining for general public.

At the moment it consists of two tools: Chain seeker and Groups seeker.

Chain seeker searches for a path between two given users. It searches through lists of buddies (and buddies of buddies and so on) to find a connection, a path between two selected people. The Small World Experiment conducted by Stanley Milgram revealed that we all live in a small social networks and that paths between us are shorter that we would expect. Chain seeker proves that the world is actually a bit smaller. The reason for this is probably that searching for a path in a digital database is cheaper than doing the same in real world.

The idea is pictured below:

Picture 1.	A chain of connections leading from person "Z" to person "A" (top-bottom approach) or person "A" to person "Z" (bottom-top approach).

Groups seeker is designed to analyze buddies list of a given person and produce a report about groups that person lives in. The tool shows that (and how) one's social world is scattered; it also tries to identify what glues the groups (is it similar age, same city, school, family?). The algorithm idea: if a buddy from my buddies list has a number of people in his list that are common for both of us, and if everyone in that group of shared buddies knows each other, then we all are a group of friends.

The idea is pictured below in its simplified form (since it omits group uniqueness and group openness factors). For the root person two groups were found. There is also one person outside of any group:

Picture 2.	Simple groups structure for root person: two groups found, one person outside of groups.

Real-life cases are never that simple. Groups are larger and they vary greatly in sizes, repetitions happen, more people are found to be outside of any group, etc.

Just a little bit more realistic example is shown below:

Picture 3.	More complicated groups structure: three groups found, one person ouside of groups, one repetition (person "E").

Before releasing the code I would like to add:

• Link nature detector: show characteristic features of two people.
• Advanced group nature detector: show characteristic features of a group of people.
• Google MAPS handling: show geographical spatiality of a chain or of a group of people.
• PostgreSQL database handling: make working with large amounts of data faster than pickles can provide.
• Distributed database system: let researchers share their data.
• A number of smaller features: make Dots Connector an easy-to-use, flexible tool.

The final aim of DC is to stop saying things about single persons. Sociologists are interested in researching society in general. I think that it is important to emphasize the direction of further development: concentration on averages, hiding peoples' privacy in statistics, interest in general view.

Some ideas for further development and research are as follows:

1. Carry out a research using Chain seeker to find out what kind of relations between people in chains are the most common. Basing on the results implement a second algorithm for buddies searching. The one implemented already is blind. The purpose of the second algorithm is to act more like humans do. With that done, conduct an algorithm war research.

   Some of the questions are as follows:

   • Which algorithm need more steps to find a chain?
   • Which need to analyze more data?
   • Is learning computers to 'think' in human categories reasonable?
   
   
2. Develop Antipathies seeker.
   

   An extension of Groups seeker tool. Its aim is to create a tool for sociologists interested in internal security.

   Basic algorithm is:

   • Find a group of X people.
   • Find somebody who knows X-1 people from that group (one connection lacking).
   • Define the lack of connection between that somebody and the person from the group as antipathy.
   • Search for the next person/go to the next group.

   Simplified visualizuation:

   Picture 4.	An illustration of basic assumption for antipathies-seeking algorithm.

   
   
   

   I have conducted an initial research among groups of my friends using Groups seeker to check if my assumption is right. Correctness of the scores was higher than I could expect.

   Antipathies seeker can answer such questions as:

   • How are antipathies influencing the shape of a group?
   • How are antipathies influencing the number of groups found?
   • Is there any correlation between the number of antipathies and the size of a group?
   • What is the average number of antipathies for a group of given size?
   • Are there any groups free from antipathies? If so: what kind of group is it? Why Antipathies seeker couldn't find any antipathies for this particular group?

   Simple modification of Antipathies seeker (for a group) may lead to creation of an individual's antipathies seeker. Such a tool reveals important privacy threat that is not generally realised: by showing your list of friends you also show your list of antipathies (although, this information is hidden in thousands of nodes connecting people within a social network service).
   

3. Develop Connectors seeker.
   

   The idea of connectors comes from Malcolm Gladwell's book The Tipping Point.

   Sociologists, along with other social scientists, strive to unveil the structure of the society and the mechanisms of social change. Their findings are broadly used by media, marketing and politics.

   Have you ever wondered why is there an infoline number on a soap wrapper? It is there because there are people among us (called mavens), who really care about the product they buy and use, and they want to know as much about a given product as possible (even if it is a soap). This kind of people is also responsible for an unofficial promotion of a product. Thus, answering their questions, and making them happy with a product leads to sales increase.

   The aim of Connectors seeker is to seek for connectors not mavens (as mavens usually unfold themselves without additional searching). I guess that this tool could be used by advertising companies (if they are not secretly using similar tools already) and this is why I would like to make it public. I think that people should know how the traces they leave in the Internet can be used by others to manipulate their actions.

   Picture below presents the simplest possibility (because the only factor taken into consideration is a connection or lack of it) in which for a connector there are three large and distinct social networks found:

   
   

   Picture 5.	A simple three-groups connector case. Based only on the structure of connections.

   
   
   
4. Add networks of groups analyzer to Groups seeker.

   At the moment Groups lists groups of friends found for particular person. It is possible to analyze the scores with pen and paper (who connects one group with another? how many connections there are between two groups? how many networks of groups have no connection with other networks of groups?) to draw a simple web of groups for particular person, but it obviously is inefficient. With this feature included Groups will be able to find the links between all the groups found and print a graph for them.

   The idea is pictured below:

   
   

   Picture 6.	A network of five groups. A connection is marked with an arrow only if there is more than one connection between two groups.

   
   
   
5. Develop a database creation tool for non-digital data.
   

   Social network services are very important source of knowledge about social networks (they are, in fact, a huge effort of millions of people to mirror the web of world societies). But, in case somebody wants to work on self-provided data, I would like to make it as simple as possible to type the data in.

   An additional feature to this tool will be a random database creator (based on a set of conditions) to support simulations.