This episode covers my presentation at Nerd Night about crowdsourcing and the Urban Voice Cambodia webmap that we have developed. This lead to a great opportunity for our NGO to partner with another in gathering spatial data in Phnom Penh. I also went to my first Khmer wedding, a sight to behold. At work I have spent much time on further checking of our datasets. A structure for the coming eviction analysis maps publication was also devised. Last, preparations began for designing the GIS directory on the server.

I'm starting to settle in at Sahmakum Teang Tnaut (STT). Through this process of making the eviction analysis maps I have had the opportunity to work with most of the staff at various points, thereby getting to know them a bit. I feel I have gotten to the point where I am comfortable relating to them and they seem to appreciate my jokes most of the time... Also, having responsibility for these maps has meant I have developed some familiarity with evictions in Phnom Penh, a focus of the advocacy work at STT since it started in 2005. I am now looking, however, to spread the responsibility somewhat. This is because I have not been given trustworthy datasets with which to make the maps. In the last update I spoke of the uncertainty over the location of some evicted communities and communities served with formal eviction notices. This fortnight involved dealing with uncertainty over the names of communities. This has caused significant stress to me as I felt under pressure to push out the maps but did not have faith in the datasets.

By talking with STT staff I have begun to regain some certainty in this aspect of the datasets. There is controversy in a number of areas about names of communities and locations:

1. There is disagreement over the Khmer language spelling of some names
2. There is a history of doing French transliteration of Khmer names, and some people still do this (such as, I think, Russei Kao)
3. Some people now do English transliteration of Khmer names (such as Russey Keo)
4. There is disagreement over what we should refer to in times of disagreement
    4.1. A LICADHO representative emphasises going with the most common/well known transliteration and/name, and often compares the number of Google hits for help judging this
    4.2. A staff member of STT stated that we should use the Khmer Government website for the correct transliteration of place names, such as this site ...for the transliterations of districts.
5. If I recall correctly, a STT staff member advised that Khmer language was simplified during the 1970s to make it easier for the Vietnamese to learn. It was stated that the government is now promoting a return to the use of more sophisticated Khmer. This transition contributes to the controversy over naming. It also means that the Khmer government's publications (such as their website) should be the standard for our names and transliterations.

One problem with the names of communities in our datasets is they sometimes contained a mixture of English and Khmer. Talking with another member of STT staff, I was advised that we should always use the Khmer name for a community. Mostly, the dataset of communities with eviction notices is like this. National Road 6A, however, had to be changed to Plouv Cheat Pram Muay A. The evicted communities dataset is also composed of Khmer names. We had to change occurances of 'village', 'street' and 'group' to 'phum', 'plouv' and 'kroum' though. Here I should mention that this has caused some disagreement for one community, Kroum 78. Khmer people will refer to it as this but LICADHO advised that it is well known amongst foreign people and NGOs as Group 78. STT management agreed with LICADHO and said to use Group 78 in this case. This highlights a separate standard for locations that are important to our advocacy work.

All of this discussion lead to my devising the following standard for names, simply a summary of what staff had said to me. I wonder if something like the following should actually go in the map publication as an appendix:

1. Official locations such as administrative boundaries (khans, sangkats), towns or lakes should be named according to Khmer Government publications (i.e. an English transliteration of more sophisticated Khmer)

2. Names of locations where we perform advocacy work should be, in order of importance:
    a) the most well recognised name amongst the media and NGOs
    b) Khmer
    c) if no name exists for the community, name it according to the closest intersection (such as Plouv 281 & 516, a community that has received a formal eviction notice) or street (such as Plouv Cheat Pram Muay A).

Speaking of standardisation, the server is now ready for us to start using. Although not quite. We need a folder structure, and set of permissions, that provides storage locations for the different teams of STT. Now that we will all be storing our data in the one location, this is also a prime opportunity to standardise our file naming and approach to organising information within each of the team folders. Thankfully, this looks like not being too difficult for three reasons. First, my experience working at Jim's Engineering 1 and Jim's Engineering 2 means that I can use their methods of structuring the server for inspiration at STT. Second, a previous volunteer gave a presentation about file naming conventions in 2009. It's a wonderfully straightforward powerpoint which I think we would benefit from viewing again as an organisation, since I have seen only one file where his suggestions seem to have been followed. Another sign of his system not yet being implemented is that, I, as a new staff member, was not made aware of the naming convention when I started. Third, everyone seems keen to discuss the structure of the server together. We all met on Friday about this, without the prompting of senior management. We got the whiteboard out and drew up STT's First Server Diagram :-)

One thing I was happy to finish last Friday was the first draft of a description of how we could structure the GIS folder on the server. I have mentioned these ideas in earlier posts, but it was good to write a formal definition for STT of the terms spatial warehouse and generic job folder, as well as describing how they are managed by administrators and used by mapping staff. It will be good to hand this over to management as my first report. I will integrate the suggestions from the earlier volunteer about file naming. Through producing this document, it could prompt the other teams to begin to think about how they will manage their portion of the server.

As you might have gathered, amongst all of this debate about standardisation and data management, little progress occurred in terms of map outputs. This does not concern me so much now that I have realised how much work was still to be done with the datasets of the maps. I did, however, make some progress in formatting the map publication. It seems most sensible that we should have an overview map of all of Phnom Penh containing the three datasets, and then district maps of a larger scale. This means we can omit labels for the datasets on the overview map, providing much more room since there won't be a large table for deciphering community names based on codes in the map. Such a format makes it much easier to gain a grasp of the basic spatial relationship between the points of the three datasets, such as where clusters match up.

This transition in presenting the data occurred after a review of the maps by an architect and designer. It's always useful to have input from other disciplines, their methods can still be applicable to one's own. A great example of this, unrelated to work, that I came across recently was with the crowdsourcing site Kaggle. Here a glaciologist used algorithms developed within his discipline for analysing remote sensing imagery of glacier mouths to analyse dark matter in space imagery. Quoting Kaggle:

"The challenge data was posted on May 23rd, 11. In less than a week, Martin O'Leary, a British PhD student in glaciology, had created an algorithm that the White House announced had 'outperformed the state-of-the-art algorithms most commonly used in astronomy for mapping dark matter.'" (para 4, source)

Why my sudden focus on crowdsourcing? Well, I presented about it at Nerd Night in Phnom Penh this fortnight. It was a lot of fun! Here's the video, it's 6:40m. Nerd Night has a special format for presentations where each has 20 slides, and the slights advance automatically after 20 seconds. I'm glad I practiced, otherwise it would have been an utter mess. In the end I thought it went well, and it served my purpose of advertising the webmap too. This is because a representative from PSE came up to me afterward who was keen to see how his NGO could get involved with contributing data to the site. We have already had one meeting and will be having another next week with some business school students from PSE, amongst whom we aim to set up a competition to encourage submissions to the site. Good that things are starting to get moving for the coming launch of the webmap on 16 February 2012.

Making Trouble in Phnom Penh

The main event that occurred outside work was attending my first Khmer wedding. It was the sister of one of my colleagues at work. I think they might have been high society in the town where it happened, the Sangkat chief attended along with some other important looking people and, most of all, 900 people showed up! It was stunning to behold how they managed the whole affair. Oh, it also went for two days. But back to the nigh on 1000 guests. They only had 55 tables of 10 chairs each. So what had to happen was families would come, eat their meal, greet the happy couple and then leave. The army of catering staff would then clean off the table (by putting all the rubbish on the floor, except the beverage cans which kids collected in a clever fashion by threading a string through the ring of the stay-on tab (I love Wikipedia!) of each of them). Then the scene would be repeated. It was quite a fantastic night with a rock band(!) on stage, 5 pretty dancers and a quite staggering amount of liquor to get through. Perhaps this is the best indication of the scale of the event: there were 30 cases (6 bottles apiece) of Chivas Regal, and 100 cases (24 cans apiece) of Angkor Draft. I certainly did my bit but paid for it with the worst hangover I have ever experienced. Don't think Khmer people can't drink! Another cause of pain the following day was the 100+ mosquito bites I suffered during the night because the net wasn't set up properly and we were sleeping outside. There certainly are more of the b*****s in the provinces.

Tomorrow the PPCTMC heads east, we aim for the Vietnam border again, on a route something like this. Next week Wednesday is also the second Phnom Penh Mapping Meetup. I've spoken to a few people who've said they'll be coming so it looks like the success of the first one might not just be a one off thing. I shall report on both next episode.
________________________

Pardon my vanity, but I'm pleased to see that on 21/1/12 this blog sits at number four in Google's search results for the string 'making maps without borders'. Curiously, there is actually an organisation going by this name! I like their south up world map on the homepage.