From ImageWeb

Demos

1. Leptospira field studies (GoogleMaps) http://maps.google.com/maps?q=http://imageweb.zoo.ox.ac.uk/pub/2008/geomashupdev/leptospirademo_time/leptospira_field_studies.kml OR http://imageweb.zoo.ox.ac.uk/pub/2008/geomashupdev/leptospirademo/
2. Time-based visualisation (Mapstraction/Googlemaps/Scriptaculous) http://imageweb.zoo.ox.ac.uk/pub/2008/geomashupdev/leptospirademo_time/

Choice of Data from PubMed

http://www.ncbi.nlm.nih.gov/pubmed/

Leptospirosis field studies <5y

• SEARCH: http://www.ncbi.nlm.nih.gov/sites/entrez?id=AdvSearch&cmd=AdvancedSearch&tab=Guided%20Search

 (("200"[Publication Date] : "3000"[Publication Date]) ) AND (leptospirosis[Title]) 

(NB searched title only - not abstract - to minimise false positives)

• RESULTS: 3504
• Selected articles with a place name in the title field
  • http://www.ncbi.nlm.nih.gov/pubmed/18541776 (Seroprevalence of leptospirosis and risk factor analysis in flood-prone rural areas in Lao PDR.)
  • http://www.ncbi.nlm.nih.gov/pubmed/18473102 (Demographic and clinical features of leptospirosis: three-year experience in central Taiwan.)
  • http://www.ncbi.nlm.nih.gov/pubmed/18447625 (Underrecognition of Leptospirosis During a Dengue Fever Outbreak in Hawaii, 2001-2002.)
• In one case (Seroprevalence of leptospirosis...) the abstract contains more specific location information (Khammouane Province, Lao) -> Search by title field only is a good way to reduce false positives, BUT then need to inspect abstract field for more accurate info?

Investigation of Technical Options

Harvesting

• PubMed XML API: http://eutils.ncbi.nlm.nih.gov/entrez/query/static/eutils_help.html
  • Example - Search for the term cancer for the entrez date from the last 60 days and retrieve the first 100 IDs and translations using the history parameter: http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?db=pubmed&term=cancer&reldate=60&datetype=edat&retmax=100&usehistory=y
  • Example - Fetch XML for an article http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=11748933&retmode=xml

Extracting Place names and dates

1. Manual
2. Calais Reuters
3. GATE? wrapper/service?
4. Others? - Various tools listed at

• • http://kmi.open.ac.uk/technologies/
  • http://ondex.sourceforge.net/pubs/skusa2005_briefings_in_bioinformatics.zip

Comparison of Calais and manual markup for 3 selected papers

 Leptospirosis, Lao PDR

 Khammouane Province, 
 Lao People's Democratic Republic, 
 December 2006

EmailAddress: leok@med.nagoya-u.ac.jp
Organization: Nagoya University School of Medicine, 
Department of International Health, PDR
Country: Japan
MedicalCondition: leptospirosis
City: Nagoya
ProvinceOrState: Khammouane Province

PDR

 leptospirosis

 Khammouane Province

 leptospirosis
 Taiwan

 central Taiwan
 October 2002 to October 2005

 Taiwan

 Fever, cough, headache, septic shock, Leptospirosis

 Hawaii
 2001-2002

 1997 through 2006
 Hawaii
 May 2001 to April 2002
 Hawaii

 chills, dengue fever, rash, headache, dengue, 
 petechiae, fever, leptospirosis

 Hawaii

 antibodies

Conclusions of comparison

1. Calais found all places except Lao, and categorised them city/province
2. Calais does not extract dates

Alternatives to Calais

GATE/Annie

• http://gate.ac.uk/ie/annie.html / http://gate.ac.uk/sale/tao/index.html#annie
• "GATE is distributed with a lightweight information extraction system, named ANNIE, which detects person and organization names, geographical locations, dates, times and money amounts. It employs a gazetteer with lists of names such as cities, countries or organizations and cue-words such as days of the week."
• "We extract place names from web pages using the GATE (or the main module ANNIE) software developed for ex- tracting named entities. GATE has been chosen instead of implementing a different method because the former has been reported to give accurate place name extraction re- sults[9]."
• TESTING:

1. Online demo - http://gate.ac.uk/annie/annie.jsp?url=http%3A%2F%2Fwww.ncbi.nlm.nih.gov%2Fpubmed%2F18541776&annotation%5B%5D=Location - finds "Lao" (better than Calais)
2. Installed locally from http://sourceforge.net/project/downloading.php and tested:

• GUI:
• Java Lib: http://gate.ac.uk/gate-examples/doc/index.html

Geocoding

GMaps

• http://code.google.com/apis/maps/documentation/services.html#Geocoding
  • REST (serversi) http://maps.google.com/maps/geo?q=1600+Amphitheatre+Parkway,+Mountain+View,+CA&output=xml&key=abcdefg
  • Javascript (clientsi) http://code.google.com/apis/maps/documentation/examples/geocoding-simple.html
• TESTING
  • 1. found 'Taiwan', 'Salvador Brazil', 'oxford', 'accra', 'timbuktu' successfully.
  • 2. local copy works

YMaps

• http://developer.yahoo.com/maps/rest/V1/geocode.html
  • REST: http://local.yahooapis.com/MapsService/V1/geocode?appid=YD-9G7bey8_JXxQP6rxl.fBFGgCdNjoDMACQA--&street=701+First+Ave&city=Sunnyvale&state=CA Fro non-USA locations, use free-text: http://local.yahooapis.com/MapsService/V1/geocode?appid=YD-9G7bey8_JXxQP6rxl.fBFGgCdNjoDMACQA--&location=Salvador,Brazil
  • Javascript -- ( But can use GeoRSS parser on client.)
• TESTING
  • found 'Salvador, Brazil', 'Taiwan', -> SAME.

Others

• http://worldkit.org/geocoder/

Map-based Visualisation

GMaps

• Google Maps API
• http://code.google.com/apis/maps/documentation/examples/
• http://code.google.com/apis/maps/documentation/reference.html
• Data extensive, eg in Salvador Brazil http://maps.google.co.uk/maps?f=q&hl=en&geocode=&q=salvador+brazil&ie=UTF8&ll=-12.970435,-38.512573&spn=0.537307,1.021042&z=11&iwloc=addr
• Demo in SVN

OpenStreetMap

• http://www.openstreetmap.org/
• Problem: data too incomplete - http://www.openstreetmap.org/?lat=-12.971&lon=-38.511&zoom=11&layers=0B0TTF

YahooMaps

• http://developer.yahoo.com/maps/ajax/
• Data completeness not bad but not as good as GMap - http://maps.yahoo.com/#mvt=m&lat=-12.996&lon=-38.494011&zoom=12&q1=salvador%2C%20brazil
• supports reading from georss http://developer.yahoo.com/maps/georss/ and automatic geocoding.

Comparison

• http://home.olemiss.edu/~jcchurch/mapapi.html
  • "Google allows you to import XML through a generic "GXmlHttp" object, but you have to parse an XML document yourself .....Yahoo allows you to import XML through a single call to "YGeoRSS" which points to your XML file."
  • " ... Yahoo has a simple geocoding service built into the API...."
  • "Google has the advantage of intergrated satellite maps, while Yahoo has the built-in geocoding service. So what do you value more? Personally, I like my satellite maps. If I need a geocoding service, I'll intergrate Yahoo's geocodeing REST API into my Google Maps application."
• http://www.sergeychernyshev.com/maps.html#-12.970382,-38.512382,6,0 - direct comparison.
  • GMaps better for salvador http://www.sergeychernyshev.com/maps.html#-12.936139547383378,-38.498497009277344,5,0
  • Yahoo better on close up sattelite view http://www.sergeychernyshev.com/maps.html#-12.936139547383378,-38.498497009277344,1,1
• Update on XML parsing - Google maps now also supports KML/GeoRSS? http://googlemapsapi.blogspot.com/2007/03/kml-and-georss-support-added-to-google.html

Tested Both

• GoogleMaps
  • KML and GEORss parsing both work in GMaps -> no diff between G and K in terms of parsing a feed. -> The automatic geocoding is now the only difference.
• YahooMaps
  • GeoRSS parsing can only read from public webservers
  • Auto-geocoding in GeoRss - does not find 'Taiwan'. "ymaps:Country string: default us The two-letter country code of the location. ca and us are currently supported."

XML Formats

• KML http://imageweb.zoo.ox.ac.uk/pub/2008/geomashupdev/visualisation/google/kml.xml
• GEORSS http://imageweb.zoo.ox.ac.uk/pub/2008/geomashupdev/visualisation/yahoo/georss.xml

Time-based visualisation?

• GoogleEarth / KML supports timespan http://code.google.com/apis/kml/documentation/time.html
• Mapstraction mesh a timeslider (scriptaculous) with google maps api http://www.mapstraction.com/demo-filters.php

Mapstraction

• http://www.mapstraction.com/ "Mapstraction additionally fills some holes each provider's current offerings (taking advantage of existing open source solutions where possible) to normalise the feature set across platforms... Features:
• Support for 9 major mapping providers
• Point, Line, Polygon support
• Image overlay
• GeoRSS and KML feed import
• Geocoding of addresses"