Video Lecture Search and Natural Language

A new speech and language search engine that could help you find particular subjects discussed in a video lecture, has been developed by MIT scientists in the Computer Science and Artificial Intelligence Laboratory. Regina Barzilay, James Glass and their colleagues say the web-based technology currently allows students and others to search hundreds of MIT lectures for key topics.

“Our goal is to develop a speech and language technology that will help educators provide structure to these video recordings, so it’s easier for students to access the material,” explains Glass. More than 200 MIT lectures are now available at http://web.sls.csail.mit.edu/lectures/ but there is no reason to think that the system could not be scaled to even the most puerile of Youtube video at some point in the future.

Meanwhile, a Cambridge, UK, company has gone into private beta with its natural language search engine. TrueKnowledge is discussed in more detail in an article entitled “What time is it at Google Headquarters?”, which is exactly the kind of question it can help you answer.

Virtualizing the Lab Book

I am a lab-note-freak who loves to write extremely detailed, organized lab notes, so organized that I really want to see the design of a really effective computer-based lab book software system.

With such a system, I’d want to be able to divide my lab work into several categories: Synthesis, Measurement, and other Manipulations. I’d also want to be able to create new categories by selecting and combining from a set of basic operations provided by the software.

Each type of lab work requires a unique form to fill in.Some fields are universal among all types of lab work such as Date, Title, Purpose, Results, Discussion, etc. But some fields depend on the type of lab work you are noting. For example Observations in the progress of a Synthesis lab work are important, but you don’t have any if you are doing NMR (a Measurement lab work) because you cannot see the sample.

With the help of database techniques your software labnotes could be searched, tagged, and, if online, shared! You could search people’s lab notes with “broadening” in Discussion field and “NMR” in Title field, for instance to learn from others’ experiences in the peak-broadening effects of an NMR study. And, of course, the digital chemical noting techniques (Smiles, InChI KEY etc.) connected to search engines could be incorporated into the software, too. I believe this is not very difficult technically speaking.

When I first heard of the online Open Notebook idea I thought it would be like the above-mentioned ideas, but now it seems that the current open notebook instances are essentially mere wikis and blogs. Wikis may be nice if you manually organize them into a labnote database but that’s much more tedious than directly using a database with a user-friendly shell. Blogs can be of some help with their datestamp format. Combined with tags you could make a blog-based lab notebook searchable, but it would still not be as good as specialized software designed for the purpose.

Perhaps I am missing the point of Open Notebook. Maybe it does encompass all my desires. I hope to learn more in the follow-up comments to this post.

— Guest blog post by PhD chemist Andrew Sun who is based in Guangzhou, southern China. You can find Andrew Sun via his Nature Networks blog where he discusses his life in chemistry.

Obesity News Epidemic

obese-overweightWe all know we’re all getting fatter, don’t we? Obesity has become the latest plague of the developed world. And, body mass index has become the vital statistic your GP is most interested.

Well, I’ve actually lost a few pounds from my Adonis-like physique* over the last few months, it must be the daily dog walking. Nevertheless, my BMI is high, but then so is that of at least half of the England rugby team – it’s big bones and muscular hypertrophy that do it. You cannot visit a health-related website or pick up a medical newsfeed these days without seeing some bizarre news related to obesity and overweight. [*Yeah, right!]

The research results are often contradictory, one day we’re told it is high saturated fat content that we must worry about. The next we hear that Gary Taubes has resurrected almost forgotten research that suggests carbohydrates are to blame for boosting insulin production and it is high insulin levels that make us fat. It sounds like a 1950s notion, too many potatoes will make you fat, but he could have a point. The link between insulin and obesity is very strong, but does one cause the other or do they operate synergistically to the detriment of our health. Who knows? Certainly not the headline writers were see, as I say apparently contradictory and at best confusing statements day in, day out.

  • Study firmly links obesity, cancer
  • Diabetes up amid rising obesity
  • Obesity ‘fuels cancer in women’
  • Obesity ‘epidemic’ turns global
  • Obesity May Be Protective in Progressive Prostate Cancer
  • Obesity and overweight linked to higher prostate cancer mortality
  • Little bit of fat not so bad: new study
  • Diabetes up amid rising obesity
  • Obesity ‘not individuals’ fault’
  • Gyms ‘little help’ in obesity
  • Inflammation, Not Obesity, Cause Of Insulin Resistance
  • Study finds some overweight people live longer
  • Little extra weight may not be bad

That’s just an almost random sample from this week’s news. But, the message is clear – we don’t really know what’s going on. The conventional wisdom has it that the more calories you take in and the fewer you use, the more overweight you will become. But, the type of calories do matter, as Taubes points out, we don’t tend to talk about middle-aged guys with burger guts, the more usual description of choice is a beer belly. The calorific content of beer, of course, arising from carbohydrates as opposed to fat.

There are also issues with the public health statements that tell us to reduce our saturated fat intake and to keep our (bad) cholesterol levels low. But, did you know there isn’t just one form of low-density lipoprotein, there are two – a dense form and a diffuse form. New evidence points to the dense form of LDL as being the bad form and not the nice fluffy type, but related research also hints that the presence of cholesterol is not actually a relevant risk factor for cardiovascular disease. It’s the dense LDL itself. So, is there any point your GP measuring your blood cholesterol and putting you on statins? Possibly not.

And, what of the possibilities that obesity is down to genetics, viral infection, bacterial infection, (fungal infection?), hormonal imbalances, pancreatic problems, missing out on breastfeeding as an infant, getting too much breast milk as an infant, a throwback to our grandparents’ diet, an evolutionary aberration, too much TV, not enough sleep, too much carbohydrate, too much protein, too much fat, too little exercise, too much walking and not enough running…

Taubes comes to 11 critical conclusions in Good Calories, Bad Calories, based on substantial literature research and interviews, summarised below:

  1. Dietary fat does not cause heart disease
  2. Carbohydrates do, because of their effect on insulin
  3. Sugars are particularly harmful
  4. Refined carbohydrates, starches, and sugars are the most likely dietary causes of cancer, Alzheimer’s Disease, and the other common chronic diseases
  5. Obesity is a disorder of excess fat accumulation, not overeating and not sedentary behaviour
  6. Consuming excess calories does not make us fatter any more than it makes a child grow taller
  7. Exercise does not make us lose excess fat; it makes us hungry
  8. We get fat because of an imbalance between hormonal regulation of fat tissue and fat metabolism.
  9. Insulin is the primary regulator of fat storage
  10. Carbohydrates make us fat by stimulating insulin secretion
  11. The fewer carbohydrates we eat, the leaner we will be

Confused? It’s enough to make you head for the donut bar. Or, maybe not. Next week, “Cardiovascular Disease News Epidemic”. Incidentally, I was going to call this post Bingo Wings and Muffin Tops, but thought better of it. You can look up definitions in the Urban Dictionary.

Scientific Intuition Under the Spotlight

Spotlight logo

It is that time of the month again. Spotlight on the physical sciences time, in conjunction with my colleagues at the intuitive portal Intute.

This month’s round-up includes the latest on the problem of stump exposure and glaciation in a globally warming world, elemental discoveries and the isotopic heavyweights, and a video interview with the researchers behind the discovery.

In the video, Dave Morrissey discusses why knowing what isotopes exist for the elements is important to our understanding of nature and how they might be made artificially in the laboratory.

Einstein Meets Hendrix

Einstein meets Hendrix

Well, not quite, but the wonderfully named Dr Mark Lewney puts on a great show not only as an axe hero extraordinaire but as a high-flying physicist who can explain why his nifty chops and runs sound the way they do. I had a quick e-chat with him the other day and we obtained permission to post his Famelab video from Channel4 on Youtube. So turn your speakers up to 11 and get ready to rock, harmonically, to the physics of heavy metal geetar!

The one thing that lets Dr Rock down is the total lack of a Justin Out of off of The Darkness jumpsuit and chest wig. Oh well, can’t have everything…

Facing up to Facebook

Facing up to Facebook

Sciencebase readers who scroll all the way down to the footer of any page on the site will most likely have spotted a clutch of new icons in a section I call Geeky Fun Stuff. I never thought of myself as an ubergeek until recently, but I guess it all adds up: big science fan, science degree, science writing as a career, fan of the more technical kinds of music, Rush, Peter Gabriel, Pink Floyd, that kind of stuff, oh and the The Dickies (I jest), and running a blog with literally thousands, well not thousands, dozens of plugins, that you spend far too much time tweaking.

And, part of being an ubergeek is stepping out of denial and facing up to one’s Facebook presence, the installation of Scrabulous, South Park character creator, and of course, the creation of one’s own niche group (science writers, 167 members and growing, by the way).

Once you’re up on Facebook, there’s no reason not to have a Twitter and a Pownce account too, as well as providing readers with direct access to your StumbleUpon and del.icio.us pages (see the footer). But, of course, those of you who reached the footer already know all this. To top it all, I guess fessing up to ubergeekicity also involves giving bits of your blog odd names, such as Elemental Discoveries, and Geeky Bits.

It probably also involves including links to a collection of my subscribed feeds known as an OPML file, worrying about how many subscribers the site has, and spending inordinate amounts of time running a science podcast that’s available on iTunes but doesn’t involve me laying down my Geordie accent in an mp3.

So, if you’ve haven’t been down under on Sciencebase, now is your chance, there is lots happening at the foot of this page. Feel free to “friend” me via any of those icons of web 2.0 but only if you want a self-professed ubergeek in your virtual circle.

The Missing Stuff of Thought

The Stuff of Thought

It would be so easy to latch on to one particular section in cunning linguist Steven Pinker’s new book, The Stuff of Thought. It’s full of expletives, cussing, and good old swear words. There’s a serious point to Pinker’s use of so many taboo words, and it interesting to say the least to hear him using them in interviews and to watch the unprepared interviewer squirm.

But, that’s as maybe. For a blog post aimed at a general audience it would be inappropriate to use some of the more savoury language discussed. Moreover, the post, and potentially the Sciencebase website itself could so easily be tagged as containing adult contact and be filtered out by search engines, proxy servers and nanny software. I could use asterisks to mask off the vowels, but you’d still know what words I was citing and it would look silly. More to the point, while the expletives, their context and usage are fascinating and certainly worthy of close scrutiny, it is the words that Pinker admits do not actually exist that are of most interest.

So what words can we discuss that don’t exist? Well, for instance, why is there no gender-free term for a single member of a herd of cattle in common usage? It sounds wrong to discuss “a cattle”, but on occasions when you do not know the sex of the bovine in question you cannot differentiate between a cow and bull.

And, what about a politically correct word to use instead of the uncomfortable his or her in a sentence where some people might use the grammatically incorrect “their” as a PC alternative. Similarly, there is apparently no emotion-free word for a heterosexual partnership as there is for married couples (spouse) or gay couples (partner). Pinker himself often feels obliged to qualify mention of his own partner as being a woman to save confusion, but suggests that lover is too romantic while, flatmate is not only too unromantic, but is seriously ambiguous, as is the word partner. I’m sure there are many more examples, but Pinker delves into what these missing words (actually the singular cattle is one of my pet peeves) tell us about language and how we think.

You can download a very entertaining interview with Pinker from The Grauniad (mp3 format, 33Mb).

Open Notebook Science

I just listened to Cameron Neylon’s fascinating talk given at Drexel a short time ago, it’s available as a podcast/mp3 via the UsefulChem Blogspot. Neylon has turned to modified blog software to help his team capture their ongoing science and is now opening his laboratory notebooks to the world.

Several things struck me from his talk. First, he points out that grad students are generally reluctant to get involved if it means more work, especially if they are not so hot on keeping a neat paper labbook, but also because their work is suddenly on show to the world. In Neylon’s field there is also the problem of tagging the materials with which his group works – short DNA sequences and proteins. Chemists, of course, have Smiles strings and InChI keys, but there is no single, simple way of tagging a protein like this, that would make it readily searchable across the blogosphere, web or database. This is especially problematic given that many research groups will be working with their own unique sequences.

However, it is the potential power of open notebook science that came across most strongly in Neylon’s talk and it is exemplified by a little anecdote he told in response to a question from the audience at the end of the lecture.

Apparently, one of his students had been struggling with a DNA experiment, finding the heatshock process difficult and not getting the results she expected. Nothing was awry in her procedures until she ran out of sample tubes and Neylon pointed out that the shelves needed restocking. It was at this point that the he and the student realised she had been using a different brand in her experiments to that used in the previously successful runs carried out by other team members.

Of course, the tube brand was not mentioned in anyone’s lab book, it was assumed they were generic components and so brand was irrelevant. Not so. At the scale they are working at, and with highly temperature sensitive materials, a minute difference in tube thickness and precise composition makes all the difference in heat distribution. The students experiments with the other brand failed because this was not taken into account. Industrial chemical engineers would have recognised the problem immediately, I’d assume. Anyway, switching back to the original brand have her almost instantaneous success and results are now being written up.

The point being, that in an open electronic notebook, such problems could be flagged so that group members and supervisor would be alerted. A meta tag in the experiment’s blog post SUCCESS=0,1,NULL could easily be included. Moreover, fields could be added in the equipment section to specify brands so that a failed experiment in which the wrong brand was used might be spotted and a different brand of tube, for instance, used next time. Such information would be archived and available to future generations so that similar mistakes would be circumvented.

Meanwhile, you can listen to the complete talk from Neylon here.

Asthma sufferers, don’t hold your breath

TL:DR – If you have asthma, do not fall for quackery, seek professional medical advice and adhere to the qualified recommendations for prescribed medication. By quackery I mean various therapies, crystal healing, homeopathy, chiropractic, osteopathy, acupuncture etc. None of it has any medical validity whatsoever.


As someone who developed exercise-induced bronchospasm (mild asthma) only after coming up to Cambridge in the late 1980s and having never suffered in childhood, I was rather disappointed to find myself on first one inhaler (a reliever) and then a second (preventer). UPDATE: 2020 – The GINA guidelines recommend nobody use Salbutamol these days, much better to be on a preventer with a combined reliever.

Anyway, asthma sufferers everywhere could benefit from breathing exercises that allow them to regain control of their breath, reduce wheezing and breathlessness, and in time cut down on their reliance on inhaled medication. When I mentioned these techniques to my GP during a general checkup, he confessed that before inhalers were available, breathing exercises were all that he and his fellow practitioners could prescribe for mild attacks. What goes around, comes around it seems.

Breathing exercises could be something of a breath of fresh air. Although saying that cold, fresh air is one of the triggers for an asthma episode as fellow sufferers will know.

Across the UK more than 5 million people suffer the potentially debilitating effects of asthma and many millions more around the world. Diagnosis is usually straightforward and most sufferers are prescribed one or both of two kinds of inhaler – an inhaler to reduce symptoms (Salbutamol, for instance, known as a reliever) and another to reduce the underlying inflammation in the lungs (a corticosteroid such as beclomethasone).

Learning to control one’s breath and to breathe through the nose is important for asthma sufferers and something many fail to do, especially when asleep.

Five golden rules for reducing your asthma symptoms:

  1. Breathe through your nose when you can, but never tape up your mouth
  2. Take control of your breathing
  3. Try to avoid nervous or unnecessary coughing
  4. Look after yourself in general
  5. Most importantly, use your prescribed medication properly

You are best advised to talk to your GP about the potential of breathing techniques for you and at the very least to adhere strictly to Rule 5. Whatever you do, do not abandon your medication. Recently, there has been a lot of talk about the Buteyko Method. This is based on a false premise about carbon dioxide levels in the blood being the problem. Don’t follow that route. Breathing exercises may well help you cope, but they will not cure your asthma.

Vote for Sciencebase

2007 Weblog Awards

I need your support! Sciencebase got nominated for a 2007 Weblog Award in the Science category, so it would be great if as many of you could vote for the site. This is the link – http://2007.weblogawards.org/polls/best-science-blog-1.php – be sure to pick Sciencebase. I’ve been a bit slow off the mark on this one, and reckon there are probably too many others way ahead of me now to ever catch up, unless all 2604 of today’s Sciencebase’s RSS subscribers vote for me right now and pass the message on!