Compared to the prior solr schemas in use for Drupal, there is a regression in that EM instead of STRONG is used to highlight search results.

CommentFileSizeAuthor
#1 1988090-1.patch1.14 KBpwolanin
Support from Acquia helps fund testing for Drupal Acquia logo

Comments

pwolanin’s picture

Status: Active » Needs review
FileSize
1.14 KB
cpliakas’s picture

I would not call this a regression, nor would I look at Drupal core as making informed best practice decisions for search. Looking at the search landscape, the following services and solutions use the specified markup for highlighting:

Search services

  • Google: <em> (has majority of market share)
  • Baidu: <em>
  • Bing: <strong>
  • Yahoo: <b>
  • Yandex: <b>

Reference for search engines & market share based on number of searches: http://searchengineland.com/google-worlds-most-popular-search-engine-148089

Search engine defaults

Solr: <em>
ElasticSearch: <em>
Autonomy Idol: <span class="highlight">
Endeca: Unknown
MS FAST: <c0> (Research might be wrong here)
Google CSE & SA: Assuming <em> since the service uses the markup

cpliakas’s picture

Based on the fact that the OSS search engine that is most popular with Drupal and the search service with the most market share use <em>, it should at at the very least open the door for more discussion.

drunken monkey’s picture

The Search API uses custom tags anyways (which we re-write to <strong> tags, btw), so I don't care either way. I don't think the comparison, at least with Google, is fair, though, as they style the <em> elements as bold. Therefore, if you'd change to <em>, you'd have to style that somehow as bold everywhere search results are displayed.

cpliakas’s picture

Right, understood, but we are assuming that there is no other reason why Google, Baidu, and Lucene based OSS engines use em tags. I would just like to see us ask the question as to why this is as opposed to assuming that there is no reason.

pwolanin’s picture

This *is* a regression since it changes the behavior of the apachsolr module, and was introduced by accident in the last release.

cpliakas’s picture

I conceded that it is a regression based on the fact that the behavior was changed in the last tag. My main point is still valid, though.

pwolanin’s picture

@Chris - certainly agree that this is worth re-examining in the future. I'll just blame Robert Douglass for the STRONG tag... since indeed even Solr 1.3.0 example config has EM:

http://svn.apache.org/viewvc/lucene/solr/tags/release-1.3.0/example/solr...

pwolanin’s picture

Status: Needs review » Fixed

committed

cpliakas’s picture

:-) Oh that Robert!

I'll post a separate issue which we should probably punt to the D8 versions of the modules.

Thanks!

cpliakas’s picture

Status: Fixed » Closed (fixed)

Automatically closed -- issue fixed for 2 weeks with no activity.