Projects

Coolhunting Projects

REQUIREMENTS:

• Software development (medium),
• Statistics (medium),
• Web design (medium),
• Community building (low),
• Level of difficulty (low to intermediate)

Membership to Virtual Tribes

Using the emotion tracking features of the new tribefinder tool try to identify prevalent emotions in different tribes, training the system with Twitter profiles, YouTube comments, and Reddit comments. In particular, look at YouTube comments and subreddits to compare the tribes and emotions about certain topics, for instance comparing the YouTube videos about Lego (positive emotions) with videos about "#metoo" (negative emotions).

Tracking the Revision of History

In earlier work we have found that contemporary historians have huge influence on later historical interpretation of the past, for instance the positive reputation of Julius Cesar and Augustus was cemented by Roman historian Livy. In more recent cases of revisionism, Croatian Wikipedians tried to eliminate traces of the far-right past in the Nazi era in Wikipedia, see https://en.wikipedia.org/wiki/Croatian_Wikipedia#2013_controversy_about_right-wing_bias

This project is about identifying similar cases and trace their treatment in Wikipedia. The goal is to show the evolution of historical understanding of past historical events based on events in the present. For instance, the Anti-Vaxxers base their belief that vaccination causes autism on one single paper, which was later show to be statistically insignificant. The same happens with theories like communism or capitalism, which over the course of history are seen in positive and negative light.

Building iDirector

The goal of this project is to build a virtual member of the board of a company. More precisely, the goal is to build a dashboard, based on the honest signals of collaboration, tribefinder, and moral values and foundations, and FFI, that identifies potentially fraudulent behavior of employees through their mailbox.

Based on the Enron e-mail dataset, build a system that reliably identifies fraudulent behavior of the Enron employees convicted and others at Enron that were not convicted, but nevertheless showed potentially suspicious behavior.

Predicting "cool" companies

The role of social networks has been widely appreciated in the entrepreneurial context as providing easy access to information, support and other complementary resources necessary for the creation and development of entrepreneurial enterprises.

Despite the extensive and diverse research on networking, there are still some gaps in what actually happens in a network and in the understanding of network operation. Its nature and role in influencing business performance is still limited to a broader and theoretical perspective. In line with these views, the aim of the project is to explore and improve the understanding of networks in an entrepreneurial context by highlighting the role of strong and weak social ties. The basis of this project is Granovetter's hypothesis of strong and weak ties, which, by highlighting the characteristics, formation and outcomes of strong and weak social ties, contributes to building the concept of the entrepreneur's social network. In particular, the focus will be on the innovative performance with which companies respond to market pressures, such as those arising from social rethinking in climate policy.

full proposal

Finding stock market trends with the mood in social networks and social trends

Mood has a great impact on the stock prices. Especially companies which are in a crises are susceptible for news and the mood of traders. Finding the point, when the emotions get positive again, would help to find the point, when the stock prices will rise. The challenge will be to identify these crises before they are publicly visible (black swans). A lot of researchers tried to made a forecast for the overall market. The problem is that the overall market is depending to a lot of variables which makes forecasts often unreliable. Also those forecasts are made on a daily base. Maybe a long term perspective would help to find global trends and sentiments before they are visible in the daily news. This would help to place in the market before the mainstream investors will do.

Idea 1:

Stock prices often drop at a tipping point. E.g. in ngrams you can see that the trend of fast food decreases since 1999 while the stock price of McDonalds dropped 85% until the year 2002. So around 2000 the rising trend of fast food stopped and that had a dramatic impact to companies working in those industries.

Idea 2:

Finding trends with a dataset of tweets, reddit and YouTube comments to fetch the mood and quantity of reactions to different topics over a longer period. E.g. the mood and quantity of tweets with the topic of fuel cells and electrical cars is changing over time and especially in 2019 the topic of fuel cells became more positive and the quantity was rising while a lot of companies dealing with this topic had a great performance since this time.

Prediction Projects

REQUIREMENTS:

• Software development (high),
• Statistics (high),
• Web design (low),
• Community building (low),
• Level of difficulty (high)

Honest Signals of WhatsApp

Build an addOn to Condor, that will take an individual's WhatsApp archive, and will constructs its social network in Condor. A key challenge will be to make this privacy respecting, which means that the indivdual will be able to look at her/his WhatsApp network in Condor, and only export the honest signals for later calculation and integration into e-mail networks.

Correlation between Happiness and Sales

This project investigates the correlation between the happiness of sales-employees and their sales-figures. Is there a verifiable relationship between the happiness of a person and the likelihood of selling a product/service?

In the best case, the sales power and customer satisfaction of an employee can be determined based on his happiness/mood at work. The goal is to have people doing the selling while wearing the happimeter, and train the system with their sales success as the dependent variable.

Measuring Mood and Emotions through Body Signals

Using our happimeter smartwatch body sensing system, predict individual mood based on body signals and body language, by sensors of the smartwatch, emotion detection in faces and in sound of voices, and other signals. You will have to use and learn tensorflow and other deep learning systems for this. The goal is to investigate which sort of emotions the sensor data of the smartwatch will be able to predict. For that you will need to conduct some data collection exercises with the happimeter in combination with Webcam-based face tracking, for instance during the COIN seminar virtual meetings.

Measuring the Happiness of Plants

Using latest results from the International Laboratory in Plant Neurobiology, build a plant sensing system using electric sensors which is capable of tracking the perception of plants and their communication with their environment measuring their signaling in their chemical and physical communication gateways.

Tips for groupwork and milestones