Super Personalized Search

Search engines, e.g. Google, have transformed to personalized search engines, i.e. they are learning your habits of search and can nicely predict what you are going to search, and not only what you are searching now. These algorithms usually depend on your previous search history and other smart things that can be learned from other people's searches.

I want to suggest to expand this to new dimensions (probably Google already does this, but I don't know of it J). The first rather trivial dimension is time. There is usually a temporal dynamics of the searches, i.e. if you search X there is larger probability that you'll search Y only within a narrow time window, e.g. one day. This statistics can be learned from other users. I've heard of research done on search patterns of medication and a time-lapse of search of side-effect as a means to detect side-effects via searches. In other words, if I search for drug X and a month later I search for "head-aches", if enough people repeat this search, perhaps X's side effect is a head-ache after a month. To conclude this dimension, the search should also take into consideration delays in search and narrow time windows of X-Y search correlations.

Another cute dimension to super personalize the search is outside events, e.g. weather. The search can correlate between your own personal attitudes toward weather phenomena, e.g. snow, heat-waves, etc. Thus, taken from other sites the weather at your own location, the search can be refined when you search for "coffee shops": if it's snowing it will direct you to a more hot-coffee oriented shop, whereas if it's hot, it will direct you to ice-coffee shops.

Other local and/or global events can also shape your personalized searches, as specific people react differently to events, e.g. election, holidays, etc. Correlating your search patterns on similar events and cross-referencing it with people within your search-cluster can further refine the search results when an event occur.

Finally, the search can be refined to your own current state of affairs. By accessing your own social media, e.g. Facebook, tweeter, it can refine the search for your own and your friends state. For example, if you just posted "This is not my day", your search can be refined to happier sites so as to lighten up your day. Another example is that if a friend of yours posted that she seeks something, your own searches can be refined so as to hint to a solution to your friend's problem, even though you searched for something else. This can augment both your social interaction media, as well as your search results.

Once along these lines, one can possibly think of other ways to super-personalize the search. I encourage you to think of them, implement them, open a start-up, be bought by Google and become a millionaire. A small (non-committing) comment in the blog will be appreciated.

Biological Wind/Watermills

The source of all biological energy is the sun, through the photosynthesis process. Ultimately, life on earth is Solar. However, as we try to get more "sustainable energy", we work also with other types of energy, namely, wind. The sights of those huge wind turbines are awe-inspiring, and somewhat disturbing. Can we also build molecular wind turbines?

Recently, it has been done. Nanostructures in the form of wind turbines have been constructed. However, as opposed to the biological world, these wind turbines produce electricity. This is unhelpful for living organisms (other than ourselves).

Another puzzling question arises: life has the unique capacity to exploit the resources at its disposal. It has recently been discovered that there are bacteria that can utilize gamma radiation as a source of food. Why not wind and/or water currents?

The answer cannot be that it is mechanical, since the other way is abundant: biological energy, in the form of ATP molecules, move flagella, which are kinds of whirly hairs that drive cells, e.g. sperms, to move. So the conversion from ATP to movement is ubiquitous. Why not the opposite?

A biological-energy producing apparatus, like in the mitochondria, utilizes a pump that converts H+ gradient over membrane into ATP (using ATP-synthase protein). In other words, a source of energy in the mitochondrion creates a concentration gradient, which is then converted to ATP, which is the energy currency in biology.

I propose to create a novel energy-producing nano-structure, namely, the biological watermill. The structure should have the shape of a water mill that transforms water current (water since most of life happens in aqueous environment, but wind current can also be used for floating bacteria or plants) into a concentration gradient and then, with ATP-synthase, into ATP. This structure will thus create a completely novel and unique energy source for biology.

Furthermore, with the new DNA-folding (see previous post), one can probably create the structure from biological materials only, e.g. DNA, or try, with novel computer-aided protein-folding software, design the structure as a protein. This presents a unique, and somewhat troubling opportunity to introduce a DNA-coded windmill into biology. Inserting this into bacteria and/or plants can have drastic ramifications. Perhaps it can serve as a novel energy source in starving countries.

To conclude, I suggest inventing a current-driven (wind or water) biological structure that produces ATP. It opens a completely new type of energy for living organisms. 

Auto-bots, transformers

Who wouldn't like to see auto-bots transformers in his life-time? I believe that the time is ripe for attempting some kind of a transformer robot. For those who don't know what I'm talking about (are there any like that here?), the auto-bot transformer is a robot that has (at-least) two structurally-distinct shapes: the first is that of an automotive, e.g. bumblebee, truck, sports-car, etc. while the second is that of a humanoid form.

I believe that the development of sophisticated humanoid walking robots have reached the point where they are autonomous enough and power-fed enough to actually make it possible to add to them another (super-cool) feature. Furthermore, the development of light-weight electric cars has reached the point where, if I'm not mistaken, the battery makeup of robots and such cars are not that different.

Hence, there are only two things missing from actually building such an auto-bot. The first is the transformation mechanism. Again, I believe that the state-of-the-art mechanical engineering can easily design a simple transformation, based on so many models of transformers in the toys market. Placing two wheels on the legs, two wheels on the fore-arms should suffice for the locomotion of the car, while the "folding"/transforming mechanism is a "simple" addition of several more motors.

The second, which is somewhat more challenging, is to place the driver's seat in this configuration. For a truly functional auto-bot, it should not only be autonomous as a humanoid robot and a car, but also to be able to actually drive a human in it. Safety issues are obviously a concern here.

The ultimate goal of this kind of project is the amazing scene from the last Transformer action movie, in which the auto-bot transformed from car to humanoid, during which there was a person inside, while keeping the human intact and safe.

If anyone is up to it, her place in the coolness hall of fame is guaranteed!

Species communication graph

Can different species understand one another? Can one ant specie understand the pheromones of another? Can a lioness understand the sounds of her prey? I believe that the answer to these questions is yes, at least in part. Most of the research on communication of animals has focused on specific species, or on whether humans can understand other species. Little was done in the sense of whether different species can understand, at least in part, communication signals of other species.

I propose a large-scale research into the communication graph of species. In "graph" I mean graph-theory, where nodes of the graph represent species and edges of the graph represent level of understanding. I hypothesize three types of edges:

The first, of genetically closely-related species, i.e. species that have diverged on one aspect but maintained the same signaling patterns. For example, perhaps different types of bees recognize the dance patterns of their generically-neighboring bees.

The second edge type is that of spatially related species, i.e. species that reside in the same habitat. Those can evolutionally benefit from understanding some aspects of communication of other species, e.g. direction to food sources.

The last edge type is that of predator-prey. I hypothesize that most (at least mammalian) predators understand to some extent the defensive communication signals of their prey. This has a clear evolutionary advantage.

The experiments to check these communication edges are obviously not simple. I can think of a naïve experiment where several types of communication signals of one species denote location of food for the second species and see whether the latter can learn or distinguish between them.

I believe this project can teach us a lot about communication in the wild, as well as more fundamental aspects of communication at large.

One research idea per week, ON AVERAGE

Dear Blog-readers.
Welcome back to The Research Fountain.
I apologize for the last several weeks in which I have not posted any new research ideas.
I have moved to a new country (USA), started a new job (postdoc at MIT Media Lab) and generally enjoyed the weather (-20 deg Celsius).
I should have sub-titled my blog as "One research idea per week, on average", so check out the new posts now.

For new-comers to my blog: I'm posting research ideas from my delirious mind. Ideas I cannot pursue myself, not because I do not think they're worthy, but because I cannot pursue them all. Those that I am doing myself, I'm "posting" in scientific publications.
You are more than welcome to followup on these ideas. I waver any claim to them. I would appreciate if you have any comment to be posted on the blog. Also, if you are doing research related to the blog posts, I would appreciate hearing about it.

The posts are not related, so if you are new and bored, you are welcome to check ALL previous posts. Hope you find something to your liking.

Enjoy, and never stop thinking of how to make the world a better place!