Conquering your Essay for MIT’s PhD program in Computer Science

Conquering your Essay for MIT’s PhD program in Computer Science

Indonesia Mengglobal Essay Clinic is our effort to provide tangible help for Indonesian applicants who want to study abroad. This essay clinic is not meant to showcase ‘the perfect essay’, but by analyzing other people’s essays (what works, what does not work, what’s good, what’s bad), we hope you can learn how to write an effective application essay and how to continuously improve your own essay. We also accept essay submissions. Click here to learn on how to participate!!

Note on the Essay

This essay was submitted by Gilang Hamidy for an application to PhD Program in Computer Science offered by Massachusetts Institute of Technology(MIT) in United States.

Essay Prompt

Please explain why you are a good candidate for graduate school. You should describe why you wish to attend graduate school, what you would like to study, and any research experience you have. Describe one or more accomplishments you are particularly proud of that suggest that you will succeed in your chosen area of research.

The Essay

In early days, computer was built to solve complex mathematical equation. People then began to build computer for another purpose, such as storing texts. Then a question was raised: can we solve complex human language problems, such as looking for a hidden knowledge within a textual information, using computer? I believe the plausible way to address this question is to model the human language into mathematical equation, so computer can compute and produce a result. I aim to answer this question by entering academic society, where I will learn and collaborate with its people to solve the question.

Academic environment motivated me to innovate and present an exceptional effort on my coursework. One of my projects was a game development using Java for object-oriented programming (OOP) course. We developed an adventure game with 3D-isometric graphics. Although majority of students preferred to use readily available game engine from internet, or develop a 2D game, we decided to create our own game engine from scratch. I was responsible in developing game logic flows and components. I utilized my basic knowledge about OOP to develop a modular game framework. It was very challenging to develop the game as our source of learning was only internet, but we managed to create a fully playable game. As a result, we earned the highest score and the most innovative game project recognition from our lecturer.

It was after I became a teaching assistant for Dr. Ruli Manurung, I encountered natural language processing (NLP) field. My frequent communication with him since my second year exposed me to his research interest in NLP and computational creativity. He told me about his PhD research in computational humor and its usage in helping children suffering with cerebral palsy. I was motivated by how an idea can have endless possibilities in real life application. It drove me to learn more about NLP and took NLP-related classes on campus. My growing interest in computer processing of human language encouraged me to write my bachelor thesis in this topic. Dr. Manurung considered my coursework in IR, where I developed a prototype of holiday destination search engine, and offered me to continue the development of news aggregator system proposed by his previous student. I took his offer and proposed improvements in several aspects of the system.

One function I developed for the news aggregator system was a module to extract complete text from a web page for IR processing. There were several papers I read, including from one of my lecturers, about extracting text from web pages using machine learning. I was unsatisfied with the accuracy of mechanism proposed because they relies heavily on supervised machine learning, where I had to provide training data, and the accuracy was unsatisfying. Looking for an easier alternative, I found that the system could utilizes RSS feed data as a pointer to the location of the article within a web page. Therefore, I developed a pragmatic mechanism to detect text content from web pages by using text summary retrieved from RSS feed as pointer. The system also support learning mechanism where it can store the document structure for future processing, so the system do not need to search the whole document anymore to extract the content. The accuracy of my approach was satisfactory, above 95% of news pages in Bahasa Indonesia can be extracted.

In addition to content extraction mechanism, my work included an experiment on my custom clustering algorithm to group the news in automatically identified topics. Non-negative matrix factorization which was used in previous work has a drawback on which the system had to determine the number of clusters it wanted to produce. This was unacceptable as in reality we are unable to know how many topics currently spreading in the news. Hence, I proposed an approach which group the documents by term occurrences. My proposed approach utilized cosine similarity and centroid to compute the cluster, and did not need to specify number of cluster it want to produce. The overall result of my thesis work was presented at The Fifth International MALINDO Workshop in text and language processing at National University of Singapore, December 2011.

After my graduation, I joined Accenture, a technology consulting firm, as a System Integration Analyst. I was involved in Program Management Office implementation project in a government-owned energy enterprise in Indonesia. We designed capabilities and developed information systems to manage thousands of capital projects in the company. I designed the core database structure and system functionality. I analyzed the requirements and the business processes so the system could be aligned to them before the system was developed by the developer team. One major constraint that I faced was designing a system which fit for all purposes because of the diversity of the business cases. Moreover, due to the client’s requirements, I had to deliver solutions within time constraints, which led me to create a system prototype as a quick-win solution before the development of the system. The prototyping approach helped me to identify more cases and introduce the client’s personnel to the tool we aimed to develop.

Along with my journey in professional career, I realized that many problems can be addressed by technology. In my project, for instance, I got an idea to create a sentiment analysis from reported issues. Because of my deep computer science background, ideas like that are keep appearing. However, my current environment, such as project constraint, limited myself to accomplish those ideas. Therefore, after several years of contributing to the world as a professional, I am ready to step in to the higher level as computer scientist to make those ideas into reality and make the world a better place. I am confident that I will deliver more contribution in academic environment, where the limit is only our own mind. By enrolling to graduate study, I expect to gain deeper understanding and craft my ideas.

Massachusetts Institute of Technology becomes my destination because it provides a wide range of research opportunities. MIT is the home of cutting-edge research in artificial intelligence including NLP and IR, such as the question-answering system and natural language understanding. Although machine learning is the most used approach in NLP, I am interested to incorporate the structured approach using parse tree and first-order logic to achieve that. Along with my study, I want to join Computer Science and Artificial Intelligence Laboratory (CSAIL) to focus on my research interest and collaborate with the people in the academia, including Regina Barzilay, Tommi S. Jaakkola, and Robert C. Miller. There, I would like to learn and combine theories from NLP, IR, and HCI to create a practical technology.

After graduation, I want to work in research community focusing in NLP application either on industry or academic settings. I want to be a part of the development and deliver valuable contributions in this potential field. Computer ability to understand human language can address more real life problems. For instance, having an intelligent system that can learn medical textbook and provides recommendation based on doctor’s verbal input. Furthermore, the current trend in which the usage of intelligent assistant becomes ubiquitous proved that this field is the future of computer for living.

I am sure that I have big potential which is proven by my determination, academic background, and experience in professional and academic roles. Those traits certainly will add the diversity in your academic society. Along with my application, I attached my detailed portfolio and resume for your consideration which is available online at (http://labs.heliosky.com/). I am looking forward to contributing at MIT and wearing the Grad Rat after the completion of my study.

Review from Christian Siagian

I like his sentence about using structured approach though it needs to be delved in deeper and to be added with some more references, for example the pros and cons, etc.

The language, essay, and sentence constructions are a bit stiff, but I think these are expected from candidates from foreign countries. If I were in the committee, I wouldn’t worry too much about it. However, the biggest issue is the actual accomplishments and how the applicant should be viewed at, especially to a school like MIT.

As how I see it, there could be two ways to do this:

  1. One is a listing of awards, or rankings, backed by great scores and outstanding recommendations (i.e. “one of the best student I have ever had”). The committee will say that this candidate has a higher percentage of succeeding.

  2. Another one is a person with a unique background and perspective who approaches a problem in a unique way. The potential professor who is interested in that problem can say “Wow, this is refreshing!“. Additionally the committee will think that this candidate is really smart and mature.

Unfortunately the essay, while good, does not have the jump-out-at-you effect, which is what the MIT folks are looking for.

I think the author should go with the second route. The language should be:

  • I know a specific (or even somewhat general) problem that, if solved, can help society.

  • I have a unique set of skills and background to tackle this in a new way.

  • In my previous experiences, I built this, this, and that.

  • I would like to modify/improve the state of the art technique to solve the above problem to advance the solution.

  • With the help of professor A who has developed invention B, I believe I will be able to accomplish my goals C, D, and E.
    ==> specifics really matter here




===========================================
Indonesia Mengglobal consists of a group of Indonesian students/alumni who are excited to share our study-abroad experiences. We hold a high regard for the value of our education and believe that a high-quality global education should be easily accessible to everyone in Indonesia.
Posts | Website | Facebook | Twitter