Yahoo 코리아에서 '비'를 치면 가수와 날씨 중에서 선택해서 검색하라는 결과가 상단에 나옵니다.
실제 검색해보면 가수 '비'는 '가수,앨범'등의 질의가 포함되어 확장해서 검색해주고 날씨 '비'는 기상,날씨와 같은 질의가 포함되서 검색됩니다.
이러한 질의어 확장을 '노가다'로 했는지 아니면 Clustering을 통해서 자동화했는지 궁금하네요.
여하튼 좋은 시도인 것 같네요.
# by | 2006/04/14 00:13 | 트랙백(77) | 덧글(0)
# by | 2006/04/08 14:18 | 검색 | 트랙백(26) | 덧글(0)
# by | 2006/04/01 15:12 | 트랙백(5) | 덧글(0)
# by | 2006/03/24 14:14 | 트랙백(45) | 덧글(0)
# by | 2006/03/24 12:02 | 트랙백(47) | 덧글(0)
# by | 2006/03/20 11:56 | Web 2.0 | 트랙백(53) | 덧글(0)
이는 현재 검색 엔진의 랭킹 중 상위 것들을 제외하면 실제 개별 사용자의 지각된 관련성과 많이 차이가 난다는 것입니다.
이는 같은 질의어에 대해 각각 개인별로 다른 평가를 한다는 것과 동시에 같은 질의어에 같은 의도를 가지고 있어도 그 결과를 다르게 평가한다는 것입니다.
본 논문은 이러한 문제를 해결하는 것이 개인화 검색이며 앞으로 계속 연구되어야 할 것이라고 결론을 내리고 있습니다.
너무 당연한 이야기인가요. -.-;;;
# by | 2006/03/16 14:21 | 트랙백(293) | 덧글(18)
We’ve been very happy to see the prominence of search in the classifieds market grow so quickly. We believe that search is a vital element to making online classifieds a great medium for both buyers as well as classified advertisers. Another search engine for classifieds launched yesterday. We welcome Vast to the crusade.
How many sites does Oodle reference in it’s index?
We index both ends of the long tail and point to over 50k sites.
How many listings does Oodle index?
In our current coverage area (just over 100 metros), we have 10M active listings. This only includes listings that are fresh and relevant: we keep track of all the listings we’ve seen and auto-expire old ones that are still online and exclude things that look like listings but aren't (reviews, spam, etc.). This number also doesn’t include listings currently in our index but outside of our coverage area. In next month, we’ll be rolling out millions more listings as we expand our footprint.
How old is your index?
Our index is refreshed every few minutes. From the time we first see a listing online, clean/tag/rank it, and push it out to our index is usually under 5 minutes (and we’re working hard to reduce that number).
It’s essential for a classifieds search engine index to be fresh. Unlike auctions, where you want to be the last person to respond, with classifieds you want to be the first. That’s also why we offer alerts, so people can be notified in minutes when something they are looking for becomes available.
How does your crawling technology work?
We use a hybrid approach. Some crawlers look for sites to crawl. We also have a farm of "micro-spiders" that target collections of similar sites. We also get data feeds from a large number of partners.
When will you offer an API?
This is mentioned in my previous post. We have an API that we've deployed with a few partners. It will be publicly available in the near future. If you're interested in using it in the meantime, email us at partner@oodle.com.
We also offer RSS which has always been available.
# by | 2006/03/16 11:34 | 검색 | 트랙백(10) | 덧글(0)
# by | 2006/03/14 11:37 | 검색 | 트랙백(48) | 덧글(71)
◀ 이전 페이지 다음 페이지 ▶
