Solving the "Brooklyn Problem"
7 likes1,800 views
The document discusses improving user experience through location-based search optimizations. It describes using a Solr index with geographic data and population information to reorder search results for a query like "Brooklyn" to prioritize Brooklyn, NY over other locations. Code examples are provided to load location data into Solr and implement a custom score function that considers distance to search point and population to rank cities. The approach enhanced data quality with minimal effort.
![ALL THE CODE!
<?php
$fh = fopen('worldcitiespop.txt', 'r');
$row = fgetcsv($fh); // discard headers
for($i=0;$i<100;$i++) {
$row = fgetcsv($fh);
$json = array('add'=> array('doc' =>
array(
'terms' => $row[2] . ' ' . $row[0], // city country
'population' => $row[4],
'latlong' => $row[5] . ',' . $row[6], // "latitude,longitude" gets parsed
)));
curl_post(json_encode($json));
}
}
function curl_post($json, $url='http://localhost:8983/solr/update/json') {
$ch = curl_init();
curl_setopt($ch, CURLOPT_HTTPHEADERS, array('Content-Type: application/json'));
curl_setopt($ch, CURLOPT_URL, $URL);
}
?>
Thursday, April 21, 2011](https://image.slidesharecdn.com/where2011-110421141524-phpapp02/85/Solving-the-Brooklyn-Problem-16-320.jpg)
Solving the "Brooklyn Problem"
- 1. Solving The Brooklyn Problem: Better UX through cheap hacks Kellan Elliott-McCrea, Etsy @kellan Thursday, April 21, 2011
- 2. Thursday, April 21, 2011
- 3. Thursday, April 21, 2011
- 4. Thursday, April 21, 2011
- 5. 1. UI/UX can be improved by applying ambient location information Thursday, April 21, 2011
- 6. 1. UI/UX can be improved by applying ambient location information 2. It’s easy. Thursday, April 21, 2011
- 7. The Brooklyn Problem Thursday, April 21, 2011
- 8. The Brooklyn Problem Brooklyn, New York Brooklyn, Alabama Thursday, April 21, 2011
- 9. if ($query == ‘Brooklyn’) { return ‘Brooklyn, NY’; } Thursday, April 21, 2011
- 10. Thursday, April 21, 2011
- 11. Thursday, April 21, 2011
- 12. Thursday, April 21, 2011
- 13. Thursday, April 21, 2011
- 14. Solr 3.10, Maxmind world populations dataset, bash, and an “algorithm” Thursday, April 21, 2011
- 15. Solr geo-spatial extension 3.10+ schema.xml: <fieldType name="location" class="solr.LatLonType" subFieldSuffix="_coordinate"/> .... <field name="latlong" type="location" indexed="true" stored="true" required="true" /> Thursday, April 21, 2011
- 16. ALL THE CODE! <?php $fh = fopen('worldcitiespop.txt', 'r'); $row = fgetcsv($fh); // discard headers for($i=0;$i<100;$i++) { $row = fgetcsv($fh); $json = array('add'=> array('doc' => array( 'terms' => $row[2] . ' ' . $row[0], // city country 'population' => $row[4], 'latlong' => $row[5] . ',' . $row[6], // "latitude,longitude" gets parsed ))); curl_post(json_encode($json)); } } function curl_post($json, $url='http://localhost:8983/solr/update/json') { $ch = curl_init(); curl_setopt($ch, CURLOPT_HTTPHEADERS, array('Content-Type: application/json')); curl_setopt($ch, CURLOPT_URL, $URL); } ?> Thursday, April 21, 2011
- 17. “a custom Solr scoring function to re-rank results based on social geo-relevance” Thursday, April 21, 2011
- 19. http://localhost:8393/solr/ placesuggest/select? q=b*&sfield=latlong&pt=37.7749295 ,-122.4194155&sort=div(geodist (),sqrt(sum(population,1)))%20asc Thursday, April 21, 2011
- 20. OM G CO NT http://localhost:8393/solr/ EX TS placesuggest/select? GE EN q=b*&sfield=latlong&pt=37.7749295 OC SIT O D IV ,-122.4194155&sort=div(geodist IN E PE (),sqrt(sum(population,1)))%20asc G! RS !!!! ON AL IZ ED Thursday, April 21, 2011
- 21. Done! Thursday, April 21, 2011
- 22. the tear down * Solr 3.10+ working geo-spatial extensions * 2.6 million points == 380Mb index * the data sucks, and it doesn’t matter (much) * a few days of work, 85% increase in data quality. Thursday, April 21, 2011
- 23. .Wanjialing, China Thursday, April 21, 2011
- 24. Brooklyn NY, population: 0 Thursday, April 21, 2011
- 25. So why aren’t people doing this? Thursday, April 21, 2011
- 26. Thursday, April 21, 2011
- 27. Thursday, April 21, 2011
- 28. Thursday, April 21, 2011
- 29. Solr 3.10, Maxmind world populations dataset, bash, and an “algorithm” Thursday, April 21, 2011
- 30. Thank you! http://www.etsy.com/listing/59401151/vintage-coffee-bean-bag-chairs http://www.flickr.com/photos/darn/141009324/ http://www.flickr.com/photos/21953562@N07/5539813593/ http://www.maxmind.com/app/worldcities http://www.lucidimagination.com/blog/2010/07/20/update-spatial-search-in- apache-lucene-and-solr/ http://www.etsy.com/listing/66013923/triceratops-forged-copper-mask Thursday, April 21, 2011