About me

I am currently a postdoc at the National Institutes of Health (NIH), Epidemiology and Biostatistics team, working on an exciting medical NLP project that aim to help in expediting the disability eligibility process at the Social Security Administration. We mainly focus on :

  • Mobility detection in free text due to the strong mobility factor in disability
  • Domestic care and self care detection in free text

I also work on suicidal prevention project:

  • In VA medical data, in collaboration with Stanford University and NIH
  • In social media (specifically Reddit suicide watch data), in collaboration with UMD under the supervision of Dr. Philip Resnik

In May 2018, I was featured as a guest in Women in AI podcast. If you would like to hear more about my work at NIH and in the mental health domain, please listen to episode 29.

I passed my PhD defense in December, 2017. I added my slides if you would like to know more about my dissertation.

I am a member of care4lang NLP group at the George Washington University, under the supervision of Dr. Mona Diab.

My work and current interests include the following:

  • Named Entity Recognition (NER) for low resource languages
    • Transfer learning NER from high to low resource languages without the use of parallel data
    • NER for dialectal Arabic and social media
  • NER for Modern Standard Arabic (responsible for NER component in MADAMIRA toolset, in collaboration with Columbia University
  • Language variants and dialect identification for Arabic and closely related languages
  • Study the impact of patterns in Arabic on predicting other useful information such as lemmas

Over the course of the last two years, I developed interest in mental health and computational social science, which led me to decide to continue working on clinical NLP. I worked with Dr. Philip Resnik from University of Maryland (UMD) on the following projects:

  • Identify crisis blogs in social media
    • Participated in CLPsych 2016 and 2017 ReachOut shared task where our system was ranked among the top systems
  • Framing and sympathy prediction using syntactic packaging

For more info about me, please check my CV.

Non technical note about me: it is always confusing for people to call me Aya or Ayah. I like to use the former, but the latter is my legal name; so feel free to use whichever you prefer 🙂