I’m trying to implement semantic similarity based on Normalized Google Distance and i have many problems to obtain correct data. I tried to simulate browser, get data via Google API (PHP, Java, R, Google Docs) and every time i had different results. Is there any proper way to get accurate and current data?
http://en.wikipedia.org/wiki/Normalized_Google_distance
Do you mean that you are using different Google APIs for different languages (PHP, Java, etc.) and get different results or do you even get different results when you call the same API twice?
I have no experience with the Google API, but I can imagine that this phenomenon might be related to the distributed nature of Google search engine. It might happen that different sets of web pages are considered for the calculation of NGD. Are the values you receive from the API at least close together? How much time has elapsed between the API calls?
No, i'm using the same API's, but results are different for "live" google search and search via API. I'm using this url:
http://ajax.googleapis.com/ajax/services/search/web?v=1.0&q=query
Then i'm parsing JSON response. This api is currently deprecated and i don't know how to do it correctly. Next problem is how to get the number of pages indexed by Google? Where can i find this information? I tried to calculate this by search word "the" but its only approximate value. The last problem is limit to daily queries (only 100).
Simply, you can't ... since there is no source available to retrieve total searches of a keyword
1) They do not provide in their API the same results as in Google Live.
2) The results are just an estimate of the total. If you navigate through the results pages you will see that for some queries there are actually less results than the number they show you.
3) Another point is that it is very difficult to compute NGD, you can check this thesis (chapter 4 (also 3)) as an alternative (method CBM): http://www.dbd.puc-rio.br/pergamum/tesesabertas/1012681_2014_completo.pdf
How it can apply for the normal text documents . Please provide complete applications of this.
Please see the following link for more information
https://arxiv.org/pdf/cs/0412098.pdf
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python