We are using Bing Custom Search, https://learn.microsoft.com/en-us/bing/search-apis/bing-custom-search/reference/endpoints, and passing a site: filter along with the search. The current filter goes something like this:
https://api.bing.microsoft.com/v7.0/custom/search?q=searchTerm site:(support.somesite.com/en-us/productName)-(/productName/2/ OR /productName/3/)
My understanding is this will return results for "searchTerm" in the site support.somesite.com/en-us/productName and exclude sites (urls) containing /productName/2/ or /productName/3/.
The idea is we have a search filter where customers can select specific versions of our product support documentation to search. Selecting the filter excludes other versions of product from search results (i.e. versions /productName/2/ and /productName/3/ are excluded from search results).
However, in practice we are getting search results that contain urls with the versions we hope to exclude.
What is unclear from the documentation - do these https://support.microsoft.com/en-us/topic/advanced-search-options-b92e25f1-0085-4271-bdf9-14aaea720930 "advanced search options" work with the site: keyword as we are attempting to use them above? Or do they only work with the search query?
Is there a good way to exclude particular url paths from the search?
No, Bing Custom Web Search API does not fully support Advanced Search Operators (keywords), which includes the "site:" keyword.
MS Support pointed out that there is a difference between what is supported in the q= parameter between Bing Web Search API v7 (not used in this question) and Bing Custom Web Search API v7 (used in this question).
Web Search API v7 documentation (left in image)
Custom Web Search API v7 documentation (right in image)
Related
I plugged in the Google Custom Search Engine to my MediaWiki site. It seems to work fine. However, how do I also make it search for results using partial matching? For example: when I searched for 'loft', it returned only the pages containing the whole word 'loft', but I was also looking for the pages containing 'loft' as a substring of some words, like 'createloft', 'deleteloft', 'loftstudy', etc.
Google doesn't provide such advanced search features. If you need things like per-namespace search, substring matching, regex search etc. use CirrusSearch, which is based on ElasticSearch.
My question is simple, how do I make a certain page be find-able by a specific keyword.
cse it's working fine it just don't manage to find everything he supposed to.
Google custom works like google search, manipulating results may not be possible, however, check out the synonyms tab in google.com/cse.
Say your users search for MBA you can configure it to show results for Master of Business Administration
I have many documents located on my disk and I want to build a search engine to search through them.
I know Google Desktop Search or Bing Desktop Search could do that. But I want to know if there's some SDK/API to do that so I can do some customization.
What I want to achieve, is that I can provide a document and the local search engine will return all the documents similar to it.
In general there are Lucene and Solr that can help to solve search related needs in Java (I guess you are using Java based on the tag GWT).
But I don't know how to do a search by example with these tools. I think you have to extract the relevant information of the document to construct a search based on it.
We know that there are standards that if some site implements them it will be compatible with zotero.
Is there any standard for multiple scrapping?
Which standard should be implemented by site developers in search result page?
Re-posted from the Zotero forum:
unAPI (preferred) and COinS both work with multiple items.
The other methods don't work for search results.
As an example for unAPI see e.g. inSpire:
http://inspirehep.net/search?p=citedby%3Aauthor%3Aellis+-refersto%3Aauthor%3Awitten
An example for COinS on search pages there's e.g. Harvard's new Hollis catalog:
http://hollis.harvard.edu/
I'm very interested in search engines.
Today in a talk I heard that google performs a text search, while more complex engines could rely on the use of metadata, which is apparently not so used by google.
Which is the difference between text search and metadata search?
Could you provide some links where I can go deeper on this subject?
Metadata is 100% text.
The reason why Google doesn't use it is because people tend to lie about their content (not automatically on purpose.)
Now, what Google doesn't use is the Keywords meta data tag (although they may be checking it out to see whether you're a liar...) They do use the other meta tags.
I just wrote a long list of meta tags supported by many systems. I still need to add many more, but out of those that are there the og:image and description and some others are very useful.
http://snapwebsites.org/implementation/feature-requirements/layout-feature-core/meta-tags-and-links-supported-core