Update: This paper is now published in the Journal of Physics: Complexity – https://iopscience.iop.org/article/10.1088/2632-072X/aba83d.
Update: This paper is now on ArXiv at https://arxiv.org/abs/2001.01126. I just returned from the Social Networks & Health Workshop at Duke University, where I presented my research on joining graph and language embeddings to predict individuals at risk of suicide! The presentation went really well, and it was great to see the work that other Fellows had done since last year. I highly recommend the Workshop to anyone wanting to learn more about advanced social network analyses and how they can be applied to study different health outcomes. If you want to know more, please get in touch with me! The project I discussed is work I am doing for my PhD qualifying exam. It demonstrates my competencies in key areas of computational social science, social network analysis, and social informatics. I show how both graph and document embeddings generate good predictive results of whether a Reddit user will post in SuicideWatch; however, both models return quite a few false-positives and false-negatives. When used in conjunction, on the other hand, the model returns few false-positives and false-negatives and has a F1-score above 0.90. Pearson correlations between the graph and two document embeddings (from doc2vec distributed bag of words and distributed memory models) show that each model is tracking along different information even though they have solid overlap. The graph embeddings, for example, only have approximately 0.2-0.25 correlation with the two document embeddings, and the two document embeddings have a correlation of 0.6. Very excited to talk about these results more when my paper is done and up on arXiv!
0 Comments
A few weeks ago, I gave a demo on word2vec in the Social Dynamics Laboratory. I based the notebook on the one I used while I was in the Facebook Artificial Intelligence PyTorch Scholarship Challenge, but added code to import all of the data within the notebook using some terminal commands. I've found Google Colab to be quite useful for distributing tutorials like this, but I still think Google has a long way to go to make Colab easier to work with. Importing and exporting data to your drive can be complicated. If anyone has any tips on how to do these tasks easier please share (and thanks in advance)!
If anyone wants to know more about word2vec or any other embedding/language models for language, events, social network graphs, etc., please let me know! I've worked with a range of them over the last two years (e.g., the graph embedding plot in my Research page is a metapath2vec embedding for multi-relational network graphs) and am happy to share my knowledge (and to learn about new techniques as well)! I'll be talking about embedding models and my experiences with them more. The next model I plan to put into use is the DeepInf model for evaluating social influence in dynamic graphs. I'm hoping to add text and other contextual information into the model pipeline in addition to other data related to social interactions over time. Click here to read the original word2vec paper. It's been quite a while since I updated my blog – so long, in fact, that so much has changed that I got rid of my old posts. Today, I just wanted to talk about a recent project I've begun with a few friends and with the Program of P.E.A.C.E. in Elmira, NY. Since this past winter, I have partnered with the Program to design and organize a free course on computer programing for at-risk youth in the local community. The program will run weekly in the summer and will cover the full range of basic Python concepts and practices, from using the terminal to running Python, making and testing scripts, working with a range of data structures, and function and class creation (plus more). I'm really looking forward to it! These skills will be incredibly useful for the students who participate – not only for their own future career opportunities, but also for those who wish to pursue higher education after the program. I'm also quite excited for the opportunity to work with local businesses and organization and engage in consulting work and training seminars to showcase how machine learning, natural language processing, social network analysis, and data mining and extraction techniques can help them reach their goals more effectively. I'm looking forward to talking about our adventures in the program more as time goes on, but until then, here is a brief background on what machine learning is and what it can do. Within this field, I am particularly interested in natural language processing. It's a fascinating field, so definitely click the links above to learn more about them if you don't know much about them!
P.S. – If you'd like to support our programming project, please check out our GoFundMe campaign! GoFundMe Campaign Story: Our program has over 10 years of experience working with at-risk youth in our community and helping them excel to meet new educational and occupational opportunities. This year, we are expanding our services to include free computer programming classes to our students to help train them for early college success and to broaden their future career paths. The funds you donate to this project will be used to purchase the computers and other teaching resources that students will use at our facility. Please see the paragraphs and links below for more information about our program and our new project! The Program of P.E.A.C.E. (Positive Education Always Corrects Errors) has successfully helped high-risk youth achieve educational, occupational, and professional excellence since 2009 [1]. Over the last 10 years, we have solidified partnerships with the key educational, political, and judicial institutions from the City of Elmira, including direct support from Mayor Dan Mandel, to expand our services to new areas within the community. For example, since 2012, we have worked with local schools to provide specialized curricula and development-focused after school programs for students in our educational outcomes monitoring group. Additionally, each summer, we provide basketball camps and educational tours of Washington, DC, to students at no-cost. The Program of P.E.A.C.E. also works with recently released ex-convicts and their families to resocialize and prepare them for career opportunities and social success more broadly. Lastly, we work with local food suppliers to distribute free meals to food-insecure community members each week. Overall, the mission of the Program of P.E.A.C.E. is to be a generator of opportunities within the City of Elmira and surrounding areas for at-risk and disadvantaged populations. Our vision is not only to be a key player in creating these chances where they did not exist before, but to build the foundation from which our participants can flourish – thereby helping our community prosper. Guided by this vision, our next endeavor is to create the first computer programing development and training center in Elmira, NY. The need for basic computer programming skills has exploded over the last ten years. Almost half of glassdoor’s top 50 jobs for 2019 require programming skills [2]. Moreover, many of these careers provide salaries in excess of $60,000. Today’s youth have grown surrounded by technology since birth; however, few possess such skills, and opportunities to learn them are limited. The Program of P.E.A.C.E. has collaborated with a PhD Student from Cornell University, Alex Ruch [3], to create a series of programming classes oriented toward introductory and intermediate applications. Alex will also teach these lessons in person to students within our training facility, located in downtown Elmira. Given our past successes of working with local youth, we plan to provide this program freely to our students while also organizing training events and consulting for local businesses and organizations for a fee, part of which will be used to support the growth of the student-centered training program. The funds you donate to this project will help us buy the startup resources we need to begin teaching our students. We will purchase 20 computers for students to use at our site as well as one projector and projector screen for classroom teaching and student presentations. Thank you for your support in helping us succeed in our next endeavor of creating positive social change through education and employment opportunities! Please feel free to contact us through Alex's website [3] if you have any questions or if you wish to speak with us about any other matter related to our program. Sincerely, Alex Ruch and the Program of P.E.A.C.E. [1] https://www.youtube.com/watch?v=TR4GiExFy7I [2] https://www.glassdoor.com/List/Best-Jobs-in-America-LST_KQ0,20.htm [3] https://alexruch.weebly.com/ |
|