Spring 2020
Time: Thursday 6:00PM - 8:00PM
Class Location: B1C206, Building 10, NIH Bethesda campus
This is a tentative outline for the course and may be updated throughout the quarter.
- Dr Qingyu Chen, NCBI/NLM/NIH (lead instructor) - qingyu dot chen at nih dot gov
- Dr Yifan Peng, NCBI/NLM/NIH (co-instructor) - yifan dot peng at nih dot gov
- Dr Martin Skarzynski (co-instructor) - marskar at gmail dot com
- Office Hours: Wednesdays and Fridays, 4:00 – 5:00 PM.
- Please do not hesitate to reach out to me to discuss your questions and concerns about the course policies, course material, and anything else. I will do my very best to make myself available to you via email and via the course Canvas site. However, if you feel that you are not getting adequate feedback, or if you are uncertain about any aspect of the course or my expectations, please contact me (qingyu dot chen at nih dot gov) ASAP.
Prior exposure to programming and Python is highly recommended, but not required to attend this class. We will provide a crash tutorial on Python in the first three weeks.
Between Electronic Medical Records and Electronic Health Records, PubMed, and collections of biomedical grant applications, there exist large quantities of medical information stored in databases waiting to be explored. Besides tables of numbers, medical records also contain a great amount of free-text paragraphs that are comprehensible by human readers but challenging to the computers. Text mining is an interdisciplinary area that primarily combines advances in Natural Language Processing (NLP), Information Retrieval (IR), and Machine Learning (ML) to help the computers understand human written language, and thus extract medical and clinical information from free-text records.
This class aims at introducing fundamental applications in text mining such as named entity recognition (NER), relation extraction, passage retrieval and document classification. The class is oriented towards hands-on experience with Python, popular text mining packages such as NLTK and spaCy, and popular machine learning packages such as scikit-learn.
By the end of this course you should be able to:
- Understand fundamental building blocks of text mining
- Understand fundamental building blocks of machine learning regarding text mining applications
- Develop Python functions for text processing
- Build fundamental text mining applications (hands-on experience with Python, NLTK, spaCy and scikit-learn)
Each student is encouraged to bring their laptop to each class.
Programing without a computer would be an exceptional feat.
For installing Python or Python related queries, please refer to BIOF309 Introduction to Python Programming.
There is no required textbook for this course.
We do, however, highly recommend Natural Language Processing with Python and Applied Text Analysis with Python for understanding text mining in more depth. If you are also interested in exploring Python in more details, we recommend Python for Data Science and its companion text A Whirlwind Tour of Python by Jake Vanderplas.
-
If you use Windows 10, please try to set up the Windows subsystem for Linux. If you use MacOS or Linux, you are all set.
-
GitHub is offering some free awesome resources to students, that might be of interest to you, depending on your background.
In this course, emphasis is placed on your active engagement with the course material and with the other members of the learning environment – that is, US (your teachers) and your peers. We will guide and support you through the course, but the onus will be on you to take responsibility for your learning.
The course is designed for understanding primary concepts of machine learning and text mining, and more importantly, having in-depth practice on these topics. It will require you to draw connections among these main topics; critically analyze, evaluate and reflect upon course concepts; synthesize course concepts; and apply the course concepts in novel ways. For these topics, having hands-on exprience is essential; you should be able to develop primary text mining applications on your own after completing the course. The course materials will enable you to develop progressively more advanced learning skills while obtaining feedback that will allow you to make adjustments and improvements – in other words, to become a self-regulated learner. It’s important to be self-aware and to know our learning preferences. If you are not a self-directed, active learner, then that’s ok – but this course might not be the best fit for you. Please contact me with your questions or concerns about the learning philosophy for this course.
- The course Canvas site operates according to the eastern time zone.
- The weeks in this course run Monday – Sunday.
- The course materials will be uploaded to Canvas and also in github for version control (will be cross-referenced).
- Weekly Canvas Announcements: before the class each week, I will post a weekly announcement in Canvas outlining the primary course information for that week.
- Other Canvas Announcements: In addition to the weekly announcement, I will also periodically post mid-week announcements when I have important updates and reminders to share with you. It is a good habit to check the Announcements area in Canvas whenever you access the course.
- Email From Me: In accordance with program regulations, any emails that I send to you will be sent to the email address that the program has on file for you; this will be the email address that you used when you registered for this course. Please check this email address regularly.
- Email To Me: As previously mentioned, you are strongly encouraged to contact me at qingyu dot chen at nih dot org with questions about the course policies, course material, and anything else.
- Course Start Date: February 6, 2020
- Course End Date: May 7, 2020
This course adheres to all FAES policies described in the academic catalog and student handbook, including the Academic Integrity policy listed on page 11 of the academic catalog and student handbook. Be certain that you are knowledgeable about all of the policies listed in this syllabus, in the academic catalog and student handbook, and on the FAES website. As a student in this program, you are bound by those policies.
All course materials are the property of FAES and are to be used for the student’s individual academic purpose only. Any dissemination, copying, reproducing, modification, displaying, or transmitting of any course material for any other purpose is prohibited, will be considered misconduct, and may be cause for disciplinary action. In addition, encouraging academic dishonesty by distributing information about course materials or assignments which would give an unfair advantage to others may violate the FAES Academic Integrity policy. Course materials may not be exchanged or distributed for commercial purposes, for compensation, or for any purpose other than use by students enrolled in the course. Distributions of course materials may be subject to disciplinary action.
FAES is committed to providing reasonable and appropriate accommodations to students with disabilities. Students with documented disabilities should contact Dr. Mindy Maris, Assistant Dean of Education.
Students are responsible for understanding FAES policies, procedures, and deadlines regarding dropping or withdrawing from the course or switching to audit status. Harassment: FAES adheres to the NIH’s harassment policies, which can be found at the following link: https://hr.nih.gov/working-nih/civil/statement-workplace-harassment Faculty and students in FAES courses are responsible for being familiar with the NIH’s harassment policies and adhering to them.
- Attendance in class is strongly recommended; however, we realize other commitments may occasionally prevent attendance. If you miss a class, please review the materials available at the course and keep up with activities and homework.
- If you are having difficulty organizing your workload and making a schedule for yourself, please contact me for advice and assistance.
- It is understood that illnesses, work or family obligations, and other unexpected issues may occur. If you have an issue that will cause you to be unable to submit an assignment on time or is impacting your ability to participate actively in the course, please notify me by email as early as possible so that we can work out alternative arrangements. In general, the earlier you inform me about issues, the more options we have available to us, so please do not hesitate to inform me about any issues that are impacting your ability to successfully participate in the course and complete your work. I strongly encourage you to communicate with me, and I am committed to finding solutions that will enable you to be successful in the course (within the framework of program policies). However, the longer you wait to communicate with me, the more our options will be constrained by program policies.
- If you do not make alternate arrangements with me, or if you fail to communicate with me about issues ahead of time, then you will receive a zero for assignments that are not submitted on time. I do not mean to be harsh, but I am obligated to follow program policies, and it is important that program and course policies are applied in a consistent and transparent manner. Consider the following: if you are submitting a grant proposal and you do not follow all of the instructions exactly as they are written or if you do not meet the deadline, then your proposal will be rejected. If you have grant funding and are expected to produce deliverables by a certain point and you do not communicate with the funding agency about any hardships that are preventing you from being productive, then your funding will be rescinded. Unfortunately, you will rarely receive second chances in real world scenarios, so it is important to start building good habits now.
The assessment focuses on (1) practicing python and machine learning skills and (2) hands-on experience of text mining on real-world datasets. Details are as follows.
-
Python and machine learning skills (pass or fail). Due 11:59pm, May 8th.
By the end of the semester, you must complete Python Programming Skill Track, Machine Learning Fundamentals with Python Skill Track, and Data Visualization with Python Skill Track. This will take about 55 hours in total to complete. Note that you could also do other tracks if you are interested, but they are optional. Please start the tracks as soon as possible and work towards the certificate(s) throughout the semester. This may require substantial work! Do not wait until the end of the semester!
-
Text mining on real-world datasets (optional). Due 11:59pm, May 8th.
By the end of the semester, you must also develop a text mining application on real-world datasets. You will (1) submit the codes to a [GitHub](https://github.com] repository, (2) write up a report to describe the method, and (3) present in the final week. More details will be released in Week 3.
| # | Date | Topic | Lead |
|---|---|---|---|
| 1 | 2020-02-06 | Course Intro and Python crash course part I | All Instructors |
| 2 | 2020-02-13 | Python crash course part II | Martin & Qingyu |
| 3 | 2020-02-20 | Python revision, text processing I, and project intro | Qingyu |
| 4 | 2020-02-27 | Text processing II | Yifan |
| 5 | 2020-03-05 | Intro to machine learning | Qingyu & Kyubum (guest instructor) |
| 6 | 2020-04-02 | Text mining applications I: classification | Yifan |
| 7 | 2020-04-09 | Text mining applications II: sequence labelling | Yifan |
| 8 | 2020-04-16 | Text mining applications III: text retrieval & Machine learning revision | Qingyu |
| 9 | 2020-04-23 | Intro to deep learning | Qingyu |
| 11 | 2020-04-30 | Deep learning on text mining I | Yifan |
| 12 | 2020-05-07 | Deep learning on text mining II & Course summary | Yifan & Qingyu |
Note that the classes from 2020-03-12 to 2020-03-26 were cancelled due to the covid-19 outbreak.