Profile cover photo
Profile photo
Lillian Lee
About
Lillian Lee's posts

Post has attachment
I'm almost tempted to write a paper on meme mutation just to be able to include this comic in the submission. (+Lada Adamic, +Justin Cheng,, +Chenhao Tan, I guess it's up to you.)

OK, NLP folks: people have used sequence-to-sequence models to perform abstractive text compression before; but how can it be that none of them named their system "seq2sq"?!





Post has attachment
Idly looking for patterns in +Brendan O'Connor's EMNLP data: the two most-cited EMNLP papers of all time, according to Google Scholar counts, both had two women as the first two authors. Counting by ACL Anthology non-self-citations, the top two EMNLP papers were solo-authored, by men with PhDs from Penn (that rhymes!). Three Daniels in the top ten. Of the 21 papers with a "?" in the title, 6 have to do with sentiment or online reviews. Four papers have "!" in the title, two of which have more than one "!".

https://docs.google.com/spreadsheets/d/14g6skVPY7BZ9zmDfslPCpHFmldNBChj8XYknUYDYuu4/edit#gid=1604332692

Was the WWW 2017 twitter account hacked, or are they just naturally terrifically funny?

Due to the exceptionally high number of submissions, we will introduce a few rules to make #www2017 the best www ever [...] final review scores will be normalized by the numbers of authors in the paper. #www2017

Rule#2 aka #reputationRule any author who already had 5+ www papers will be automatically accepted #www2017 (Btw this is not a new rule).

#3 the harakiri rule. Any author who withdraw his paper before Dec. 1st will be silently praised and never forgotten.

#5 the diversity rule. Papers not containing the words web, internet, data, information, and rank will be automatically accepted.

rule#7 Only p-values < 0.0001 will be considered significant as per the new Aussie English dictionary definition of significance



Post has shared content
Universities are outsourcing their functions to third parties. In this post, I make the case that this is happening at the expense of student privacy, and thus marks a dangerous trend that should be reversed.

http://hackingdistributed.com/2016/08/29/outsourcing-in-higher-ed-considered-harmful/

Check all of A-E that apply:  if you, in your opinion,  generally both write decent papers and are a decent reviewer:

A. There was a key issue where I, the author, was right and the reviewer was wrong.
B. There was a key issue where I, the author, was wrong and the reviewer was right.
C. There was a key issue where I, the reviewer, was right and the author was wrong.
D. There was a key issue where I, the reviewer, was wrong and the author was right.
E. I see what you did there.

Post has attachment
Some odd journalistic choices made in this caption and photo. They are in fact talking about me, +Chenhao Tan, +Vlad Niculae, and +Cristian Danescu-Niculescu-Mizil. h/t Vlad.

Article here, see second image: http://www.thesun.co.uk/sol/homepage/features/6925626/Use-calm-language-and-8-other-tips-to-win-any-argument.html. 
Photo

Post has attachment

Post has attachment

Title: It's Not Easy Being Green; or: You Can Have Both Great Stage Presence and Great Stage Fright, And Your Own Rosemary's Baby, To Boot.

When giving talks, I usually mention how nervous I am, sometimes at the outset, sometimes in the middle when I say something strikingly non-fluent. These talks have often gone very well, so much so that people have --- in a perfectly friendly way --- suggested that I must be faking the nerves.

I am not faking. I usually get very nervous before talks, sometimes to the point of getting physically ill, occasionally acutely so. Lest some think I'm exaggerating, let me point out that there are three talks I've canceled due to illnesses that I now recognize and/or admit were related to these nerves. CoNLL 2010 was perhaps the most spectacular one, since I managed to force myself all the way to the hotel in Sweden but couldn't make the final trek to the venue. (Honest-to-god, my heart is actually pounding a little right now just recalling these episodes.)

So, it's great that people think that I'm Clark Kent/Superman[1]: completely aware of and in control of talk-giving superpowers, but hiding behind a meek persona for ulterior reasons. It's great because it's terrific that people often love these talks. But I'm much more Bruce Banner/the Hulk, talk-giving-wise. I can become this powerful (if somewhat out-of-control[2]) personality on stage, and who doesn't want to Smash! Puny! Related work! But nobody envies Banner the transformation process.

The Banner/Hulk metaphor matches another aspect of my experience, that of "leaving my body'' as another pilot takes over. Here's John Lahr[3]: "In defense against the immobilizing terror, sufferers often split off. They disassoci­ate. They report out-of-body experiences, a sense of watching themselves go by. ("It's a negative ecstasy,'' Fry says. "Remember that 'ecstasy' means 'to stand outside.' You stand outside yourself.'' "[Stephen Fry's stage fright was notorious.]) Or as Sara Solovitch says [4]: "My fear was at the controls, like an independent organism emerging from inside me, my own Rosemary’s baby.''

Anyway, just because someone seems to be confident doesn't mean that they actually are. And, when someone who seems confident tells you that they aren't, this might not be false modesty, but actually true. For me, I try to be up-front about my nerves because getting the audience to be sympathetic and friendly helps me tamp down the panic. If, instead, I don't get the audience on my side quickly, things go downhill; those who've seen me enter my catatonic-yet-babbling state when a technical glitch occurs have gotten a glimpse of the abyss.

Now, I am not saying that you have to get nervous or worked up in order for a presentation to go well. Rather, I want to emphasize that if you get very nervous before a presentation, you can still knock it out of the park. (Figuratively.) But, some very useful coping advice I've gotten from a Cornell-supplied personal coach[5] is: if you aren't naturally good on your feet, then make up for it with extra preparation. For a new talk or lecture, things go best if I write out every line I plan to say by hand, and try to review it the night before and the morning before the talk --- this gives the ol' autopilot a map to use when it kicks in.

Finally, there's also no reason to avoid medical consultation because you're afraid it might mean you're "weak''. That's just letting fear win, and that's definitely weak.


[1] True comic-book aficianados: yes, you're right, these aren't perfect metaphors because the true or retconned backstories don't necessarily match the case I'm building up. The best response I have is something David Donoho wrote with respect to email: "I am still struggling to find a metaphor for this experience. My standards for the metaphor are high. This metaphor must guide me to into a frame of mind where I will instinctively make proper use of the technology, to live moment by moment in a way that will keep my head above water. (Well, there's the beginning of a metaphor there: a flood, a swimmer, a struggle, drowning, ... too depressing, unhelpful, bad metaphor. "Down, metaphor, Down,'' I cry silently)''.

[2] SIGDial 2014 attendees may recall my actually falling off the stage.

[3] ``Petrified: The horrors of stage fright'', The New Yorker, 2006

[4] Playing Scared: History and Memoir of Stage Fright (which I cop to not having read; I'm instead cribbing from Joan Acocela's wonderfully titled New Yorker 2015 review, ``I can't go on! What's behind stagefright?'')

[5] Awesome Cornell perk: this place runs some terrific professional development programs. Check whether your institution offers anything similar! The particular program I'm alluding to was a leadership training workshop for tenured women in the sciences.

Wait while more posts are being loaded