jdstudio
BANNED
- Dec 13, 2011
- 461
- 166
Found this interesting read: Google Scholar‘s Ranking Algorithm: An
Introductory Overview Although it is about the scholar ranking algorithm, I would imagine there are similarities with Google's web search algorithm.
Here are the summary:
1. Overall, Google Scholar‘s ranking algorithm relies heavily
on an article‘s citation count. As a result, Google Scholar
strengthens the Matthew effect and is more suitable when
searching for standard literature than gems, the latest trends,
or articles by authors advancing a different view from the
mainstream. Should Google Scholar become as popular for
academic articles as it is for websites, the ranking algorithm
will create further incentives for scholars to actively
influence, or manipulate their citation counts.
2. Google Scholar‘s ranking algorithm puts a high weighting on
words in the title. Knowing this, authors should think
carefully about the title they give their articles. All relevant
keywords should be included and it might be sensible to
choose a long title.
3. Google Scholar considers only words that are included in an
article, no synonyms. For this reason, users of Google
Scholar should perform searches not only for one keyword
but also for its synonyms. Otherwise they will miss out on
relevant documents.
4. It appears as though Google Scholar does not put a weighting
on the frequency in which search terms occur in the full text.
This means that an article will not be ranked higher for a
certain search just because the search term occurs more often
in the full text. Because of this, it could be beneficial for
authors to abstain from a strict terminology in their articles
and use more synonyms. This would make their documents
less readable but more retrievable in Google Scholar.
5. Google Scholar seems to weigh recent articles stronger than
older articles which in turn, could compensate for the
Matthew Effect.
6. Google Scholar is not indexing text embedded via images.
Authors should avoid inserting tables, diagrams and figures
as images (.png, .gif, .jpg, etc.) but use vector graphics and
real text instead.
7. Google Scholar uses different ranking algorithms for a
keyword search in the full text, keyword search title, the
‗related articles‘ function and the ‗cited by‘ function.
8. Google Scholar‘s ranking algorithm puts a high weighting on
author and journal names. Users should be aware of this
because it could distort the result list. In addition, it could
also be beneficial for an author to publish in a journal whose
name includes keywords relevant to the article‘s content. The
impact of an authors‘ and journals‘ reputation on the ranking
has not been researched yet.
The research undertaken can only be seen as a first step. Many
questions have remained unanswered. Further research is required,
with more sample data and more analysis in order to get a
comprehensive picture of Google Scholar‘s ranking algorithm.
The paper can be found here:http://sciplore.org/wp-content/papercite-data/pdf/beel09.pdf
Introductory Overview Although it is about the scholar ranking algorithm, I would imagine there are similarities with Google's web search algorithm.
Here are the summary:
1. Overall, Google Scholar‘s ranking algorithm relies heavily
on an article‘s citation count. As a result, Google Scholar
strengthens the Matthew effect and is more suitable when
searching for standard literature than gems, the latest trends,
or articles by authors advancing a different view from the
mainstream. Should Google Scholar become as popular for
academic articles as it is for websites, the ranking algorithm
will create further incentives for scholars to actively
influence, or manipulate their citation counts.
2. Google Scholar‘s ranking algorithm puts a high weighting on
words in the title. Knowing this, authors should think
carefully about the title they give their articles. All relevant
keywords should be included and it might be sensible to
choose a long title.
3. Google Scholar considers only words that are included in an
article, no synonyms. For this reason, users of Google
Scholar should perform searches not only for one keyword
but also for its synonyms. Otherwise they will miss out on
relevant documents.
4. It appears as though Google Scholar does not put a weighting
on the frequency in which search terms occur in the full text.
This means that an article will not be ranked higher for a
certain search just because the search term occurs more often
in the full text. Because of this, it could be beneficial for
authors to abstain from a strict terminology in their articles
and use more synonyms. This would make their documents
less readable but more retrievable in Google Scholar.
5. Google Scholar seems to weigh recent articles stronger than
older articles which in turn, could compensate for the
Matthew Effect.
6. Google Scholar is not indexing text embedded via images.
Authors should avoid inserting tables, diagrams and figures
as images (.png, .gif, .jpg, etc.) but use vector graphics and
real text instead.
7. Google Scholar uses different ranking algorithms for a
keyword search in the full text, keyword search title, the
‗related articles‘ function and the ‗cited by‘ function.
8. Google Scholar‘s ranking algorithm puts a high weighting on
author and journal names. Users should be aware of this
because it could distort the result list. In addition, it could
also be beneficial for an author to publish in a journal whose
name includes keywords relevant to the article‘s content. The
impact of an authors‘ and journals‘ reputation on the ranking
has not been researched yet.
The research undertaken can only be seen as a first step. Many
questions have remained unanswered. Further research is required,
with more sample data and more analysis in order to get a
comprehensive picture of Google Scholar‘s ranking algorithm.
The paper can be found here:http://sciplore.org/wp-content/papercite-data/pdf/beel09.pdf
Last edited:
