11:35 - Dr. Keselj is introducing the speaker.

11:36 - Studied at Acadia, work mostly from his thesis at U. of Alberta. Trying to put linguistics back into machine translation. Adding syntactic constraints.

11:37 - Outline: tutorial on statistical MT, word order, syntactic cohesion. Phrasal SMT - a decoder from French to English and vice versa.

11:39 - Rule based translation is very common, e.g., Babble Fish in Altavista and Yahoo. Hard to fix problems without introducing new ones. A different approach is treat MT as a math problem: trans(e) = argmax_f [ P(f|e)]

11:41 - f and e are not well defined. Ridiculous: search over all French sentences. Instead: find all possible French sentence that can be created by breaking down and reassembling an English sentence.

11:42 - Need a score to evaluate the different possible sentences.

11:43 - Exampe: begins -> debute. Existence of big collections in English and their translation in French. Available free online. Sentence alignment is easy. Word alignment within a sentence is harder, but we assume it works.

11:46 - tri-gram and 4-gram models for language modelling. No need for bilingual text to build these language models.

11:47 - Word order which to translate first affects quality. Reordering a sentence is same as travelling salesman. Factorial complexity. However, language has structure so we don’t need to examine all combinations.

11:49 - Represent a sentence as a dependency tree of which word describes which. Uses CFG. Subtrees are always continuous substrings (phrases). [basically POS parsing]

11:51 - assume that subtrees are units of thought and are language universal, i.e., preserved during translation across languages. This is called cohesion. Can be verified with word alignment in parallel text.

11:53 - project tree from one language to text in another language. If subtrees don’t intersect then we’re good.

11:54 - Fox (2002), used word alignment to validate cohesion, 90% of French-English holds. Both sides win: stats say 90% is not 100%, can’t do much; linguists say 90% is better than 80%, and it’s almost 100%.

11:56 - Google and Microsoft use phrasal MT, winning approach until last year. Similar to example-based MT. Idea: trust what you’ve seen people do before over any computed translation. Phrase or expression translation (hold your horses) vs. translation in context (the voting session begins tomorrow).

11:59 - Phrase tables from aligned text + statistics from aligned text (maximum likelihood of P(e|f) and P(f|e) )

12:00 - Translation process. Load all your options (different translation for each word or phrase). Translation is similar to single-player search.

12:02 - Use beam search, one of the favourite NLP tools.

12:03 - scoring states. A linear update that adds a weighted probability ( lambda log P(e|f) for each word)

12:05 - weights are determined by training to maximize match to some human translator. (BLEU metric). Precision with respect to a reference translation. (word count)

12:07 - Modifying the search to account for cohesiveness. Questions: is that translation cohesive? Not based on POS parsing or tree but on a notion of phrases.

12:10 - subtrees don’t match phrases because phrases cross some arcs in the tree.

12:12 - if intersection happens at a boundary point then we don’t care. What is a cohesion violation then? Should push the process to the decoder and send the decoder in a different direction. Detect if a subtranslation is not cohesive and do something about it.

12:15 - Shows an example of a ‘decoder-eye view’. At the end of the day, eliminating interruption in translating subtrees is equiv to eliminating violations.

12:18 - use soft and hard constraints. Hard: if you detect an interruption through it away. Since only 90% accurate, use soft constraints. Use interruption count as a new feature in the log linear score. Constraints change when you can use a phrase, not which phrase you can use.

12:20 - Lexical reordering: a data driven model by examining the bi-grams.

12:21 - Experiment: English to French translator, built by extending an open source system.

12:23 - in base line, 4/5 are cohesive already, with high BLEU score. The other 1/5 have lower BLEU score, maybe means a difficult translation.

12:25 - Overall, base < lex < coh < coh+lex, where base << lex in 1/5 with violations and insignificant elsewhere.

12:26 - results evaluated with human evaluators. They agree little except when coh helps. System produces more literal translations.

12:28 - conclusion: method to move blocks around as units but then translate them separately. Can be applied to named entity, collocation, etc. Non-cohesive translation means it’s a hard one for phrase-based systems.

12:29 - Six years in PhD. Open for questions.

12:30 - Mike McAllister: how do you account for adjective location differences across languages. Answer: doesn’t play with statistics and doesn’t work well with text-based approaches. Current system can be model as a finite state machine. Need a more powerful model such as CFG.

12:31 - Question: why the 10% exist? Answer: 5% come from ‘not’ -> ‘ne pas’, others include heavy paraphrasing in human translations, NLP parsers are imperfect. Question: how will it work with German where verbs move to the end? Answer: doesn’t work for German. BTW, never seen english parse trees, would be interesting to see the parse tree for ‘ne pas’.

12:37 - Question: what’s the effect of punctuation? Answer: unreliable. Can improve performance a little. Hard to train because human translators are creative. Can help by restricting moving words across commas.

12:40 - Question: how to handle unseen words? Answer: There are approaches based on modelling new words, currently just copy as is.

12:41 - Vlado Keselj: how to handle source language that are not cohesive? Answer: can’t do.

12:43 - Tony Abou-Assaleh: Google won competitions up to last year, who is winning now? Answer:

• Chinese to English: MS Asia
• Arabic to English: Google
• Unlimited Data: Google
• Microsoft approch: combine many systems on closed data.
• ISI: generally good.

… a couple other questions, but blogging stops here.

Once you have a blog on posterous, you can simply and easily change the subdomain name, title, and other site and author info. None of the info are required - fill out as much or as little as you like. It’d be interesting to see who takes over the service first: real users or spammers.

TechCrunch has more info about the company and how it compares with similar services.

1. Backup Current Installation. Sounds pretty simple, but backing up from a hosted server, although not complicated, is not that simple either. You need to backup all your files and you need to backup all the database tables. God forbid you have more than just the WordPress installation using the same database. Trying to go through all the tables and deciding which belong to WP and which belong to other applications is unpleasant at best.
2. Deactivate Plugins. No complaints here. Fairly simple.
3. Download Latest WordPress. Granted, there is nothing simpler to do, save for scratching your head.
4. Replace WordPress Files. I have a few custom plugins, custom themes, and some other modes. How do you replace only what needs to be replaced but not the other files? The naive way is to do it one file at a time. You can try doing it in a batch as in copy-and-pray. Actually that won’t work because you need to be careful not replace your wp-config.php file. Further, you need to copy the contents of the wp-content folder from your previous installation, but not the entire folder. This step is what kept me from upgrading earlier - too many things could go wrong. After replacing all the files, you need to visit the upgrade page so the database schema can be changed to the new one.
5. Reactivate Plugins. You have to reactivate your plugins one at a time. There are no “plugin profiles” in WordPress, which would’ve allowed you to enable and disable plugins in groups.

In a an article mentioned in a recent Search Engine Guide newsletter, I read about this super useful almost working plugin called Wordpress Automatic upgrade. It does what it says. I chose the semi-automatic method where I clicked “next” after every step to make sure all went smoothly. And it did, except for the last step. The plugin failed to re-enable the previously active plugins. A small incovenience compared to all the other things that went right: created backup, disabled plugins, downloaded latest WordPress, replaced all the right files and kept what should be kept.

Unfortunately, the copy of the K2 theme that I had was incompatible with WordPress 2.5. Particularly, the panel manager of K2 didn’t show up in the admin section. After a bit of poking around, I downloaded K2 candidate release 6 which worked like a charm after some minor tweaks. In the process, I learned a bit more about actions, created a new plugin to customize my K2 TAA theme, and made use of K2’s action hooks to add a custom footer. My next K2 upgrade - fingers crossed - will not require changing any of the K2 code.

The code of my K2 TAA Theme plugin is reproduce below. Feel free to use it as a basis for build your own customizations for K2 and other WordPress actions.


/**
* Plugin Name: K2 TAA Theme
* Plugin URI: http://tony.abou-assaleh.net
* Description: Support functions for the K2 TAA Theme
* Author: Tony Abou-Assaleh
* Author URI: http://tony.abou-assaleh.net
* Version: 1.0
*/
function k2_taa_footer () {
?>
<p>
Copyright © 1997-2008 Tony Abou-Assaleh, Halifax, Nova Scotia, Canada.<br />
Verbatim copying and distribution of this entire article are permitted worldwide, without royalty, in any medium, provided this notice, and the copyright notice, are preserved.
</p>
<?php
} // k2_taa_footer
// Action calls for all functions
add_action('template_footer','k2_taa_footer');
?>

There are still a few lines of text missing from the thesis and a new collectible Perks Coffee Gold Card will be started soon. Stay tuned for a report on another full coffee card.

I wouldn’t do for money what I wouldn’t do for free.

Ever before and ever since, this principle has been my guiding principle in all that I do for money. I wouldn’t take on a project if it wasn’t interesting enough that I wouldn’t mind doing it for free. I follow this principle in my consulting work, in my employment, and in my volunteering work. But only in the latest that I actually do it for free.

This principle is about loving what you do and enjoying it beyond the material reward. It is about contributing to the project and wanting it to succeed because it is interesting and valuable, not because there is a pay cheque waiting for you. And it is about selecting projects that provide you with additional benefits and rejecting projects of limited personal and professional value.

This principle does not imply in any way that one should work on the project for free. There are enough interesting projects in the world that are also worth money. Money is an important element in our everyday life and at least some of the projects one works on must have a material payout. In that sense, this principle encourages you to seek financial return for your contributions while freeing you from limiting your contributions to the exact amount of the payment. For example, you accept a project that involves developing a software component for a client. Say you have a contract with the client to pay you $2,000 for the project based on your estimate of 20 hours of work. After working on the program for 15 hours, you realize that you will not be able to complete the project to your liking during the remaining 20 hours. At this point, you have a choice to make: contribute extra hours and complete the project even though you will not be paid for the additional hours, or do less (though still within the contract specification) so you only use up the remaining 5 hours. If you followed the principle “Don’t Do For Money What You Wouldn’t Do For Free”, then this project would bring more value to you than the monetary amount. Consequently, you would choose to put extra hours, perfect the deliverable for the client, gain the added satisfaction, and build a reputation for exceeding the requirements.

As the example above illustrates, selecting tasks that are interesting, challenging, and provide a learning experience enables you to succeed in your work. You are more likely to succeed in this case because you have more freedom within the time and budgetary constraints imposed by the client. You are more likely to succeed because your gain from project - new experience, new skill, expanded portfolio, foot-in-the-door with client, new connections, etc … - makes this project a win-win situation whether you get paid for all of your effort or not.

and brightness of the night

Fairy tales of truth

of the horseless knight

Sleeping before noon

waking by midnight

Dreaming of the life

sleeping in the light

Silence lost in screams

of this peaceful fight

– Tony Abou-Assaleh, May 22, 2008, Halifax, NS.

• Map caching has been optimized, providing faster map load time
• The map projections has been switched Mercator, the industry standard for online interactive maps. The streets now look more natural at zoomed-in levels.
• All Empire Theatre listings in Canada now include showtimes. Try a search for showtimes in Toronto, or movies in Halifax to get a list of the Empire Theatres in the city, along with other relevant results.
• All user contributed business listings and updates, as well as suggested related web sites have been included in this release.
• Web 2.0 fever has you? Feel compelled to share a business or a results page with family and friends? It is easy to do so using the ’share’ menu. You can post pages on Facebook, MySpace, and StumbleUpon just to name a few.

In a recent media announcement of this release, I was quoted saying:

If there is a feature that you’d love to see on our Local but we don’t have it yet, let us know through the feedback link on the website. We are committed to improving our platform to better service the end users.

And it is true! I take user feedback very seriously and read every single one of them. So make sure you let me know what you’d like to see in the next release. Who knows, the Genie might grant your wishes.

When Stephen Noton (Director of Search Engine Optimization with GenieKnows.com) registered himself and his 2008 smart car into the ‘One Lap of America’ race, he didn’t plan on telling anyone. “I have always be involved in cars and racing and the Cannonball was something that I wanted to do for many years” said Noton from the company offices in Halifax, Nova Scotia.

It wasn’t until he requested time off work that colleagues became aware of the event and Stephen’s involvement. News spread quickly around the office that Stephen was looking for a co-driver for the race. “There was an immediate buzz when the guys at work found out I was looking for a co-driver,” and Tony Abou-Assaleh (Director of Research at GenieKnows.com) quickly volunteered.

With two of their key employees involved in the event, it was hard for company executives to ignore the enthusiasm. “Once we understood the history, culture and nature of the ‘Cannonball One Lap of America’ race, we immediately saw the potential of promoting our new Local Search Engine,” said Barbara Miller Manning, President and CEO of GenieKnows.com. “Having Stephen involved in this event, and entering his smart car, really tied in well with our corporate mantra of “the little engine that could,” said Manning.

After some talks with event organizer, Brock Jr., the company not only decided to sponsor Stephen’s personal entry, but also to support the entire event as the Digital Media Sponsor.

“GenieKnows.com wanted to bring the local and community aspect to the One Lap of America event and decided to help start the OneLapper.com community website,” said Noton. This is a site where ‘One Lappers’ can get hosted blogs, upload their One Lap videos and photos and interact with the worldwide web community via this “Web 2.0″ platform.

GenieKnows.com is a technology provider servicing millions of search engine users monthly via its focused web-search portals like GenieKnows Local, Games & Health. “This is a perfect way to showcase the GenieKnows Local search product to the One Lappers as a way to search, map and find services during this year’s event.”

The GenieKnows Local search team will be setting up special entry web pages for each of the venues during the 8-day event, which will give participants the quickest way to find that repair shop, gas station, or even the track if they get lost!

Stephen Noton’s 451 smart fortwo, with an engine size of 999cc, is the smallest entry in this year’s event and just like GenieKnows.com, it is the little engine that could. Follow the race online at http://www.onelapper.com and search like a local at GenieKnows Local - http://local.genieknows.com

Mark Harper,
VP Strategic Relations

For comments or feedback please email local.feedback@genieknows.com.

Related Sites:

• One Lap of America Official Website: http://onelapofamerica.com
• Project Car forfiveone Page: http://www.451s.com/profile/forfiveone
• One Lapper - One Lap of America Community Site: http://onelapper.com
• GenieKnows Local Search: http://local.genieknows.com
• GenieKnows Games Portal: http://games.genieknows.com
• GenieKnows Health Search: http://health.genieknows.com

I invite you to join the event in person and show your support for the cause by sponsoring me. I suggest a minimum donation of $5, but if you donate $20 or more then you will get a tax receipt.

Event Details:

• Title: Revolution 2008: Spinning for Life!
• Location: Historic Properties, 1869 Upper Water Street, Halifax, NS
• Date: April 18, 2008
• Time: 10:00 AM - 6:00 PM
• GenieWheelies Spinning Time: 5:00 PM - 5:30 PM

My personal goal is to raise at least $100 and I ask for your help. To donate online, please visit:

http://my.e2rm.com/personalPage.aspx?SID=1764952

or send me a message if you prefer to make a cash or cheque donation.

Thank you for your support!

Update 2008-04-18: Spinning with the Genie Wheelies and other participants was a lot of fun. They must’ve had at least 100 stationary bikes there. The coaches were very energetic and up beat, and the music was lively and pumping. Collectively, our team of 7 raised $585 from 22 sponsors, exceeding our goal of $500. My personal thanks go Hathai Tanta-ngai and Kate Patterson for sponsoring me.

This car is awesome. It is so awesome that I thought it was manufactured like that. And not until the owner/dirver/creator started pointing out all the different mods that I began to realize how special that ride was.
So much leg room and head room, way more than most economy-sized cars that I drove or rode in over the years. The custom GPS looks as if it was factory installed. The mini LCD displays make it look that much more geeky and sporty at the same time.

I loved the sport shocks going on Halifax streets after a rough winter (well, it still is going, may even get worse). And though I didn’t get to used the add-on arm rest - it was reserved by the driver - I totally appreciated the added comfort the driver was getting.

Only one pickiness I can think of: what’s up with the doors requiring a double slam to close? Makes you want to kick the door every time it doesn’t close.

The original incarnation of this articles appeared at 451s.com