Telling a computer what to search for is a complicated business. Many of you are used to the SPIRES syntax with all its power and quirks. We’ve worked hard to retain the search syntax you are familiar with as well as allowing for a new Google-style search syntax.
What you might not realize about SPIRES syntax is that is actually the database query language. When you search SPIRES, you are searching the raw data. The language is actually incredibly powerful and only a few people even know all the possibilities it has. But because of that power, there are many ways to search and different people become fond of different ways.
Moreover, a search system must interpret the values you type in your search queries. It would not be reasonable to expect you to type query strings with raw data exactly as it occurs in the database. When you search SPIRES, the search system `massages’ the values you enter, trying to recognize various date formats, or incomplete article numbers, or trying to tell author first names from family names. INSPIRE uses different search technology than SPIRES also in this respect. To reproduce SPIRES search experience in INSPIRE therefore means not only to support SPIRES search syntax per se, but also to perform the SPIRES style of query value massaging, again with all its power and its quirks.
Most individuals, we find, use a small subset of functions of the SPIRES database query language, but different people tend to use different subsets.
That means we had to built in a translator for SPIRES query language into INSPIRE. Unfortunately, that is a daunting task with limited returns at some point. We know that some you use some fairly obscure feature of SPIRES that is rarely used overall and that would be extremely demanding to port to INSPIRE so we might not be able to provide every single search syntax that you use. However, we have scoured our search logs to ensure that a very high fraction of you will be able to use identical syntax as you transition from SPIRES to INSPIRE.
We urge you to try out the new INSPIRE syntax as well as you’ll find it is just as powerful and has extra features because of how the underlying Invenio engine uses the ideas of papers being described by logical fields and their values, as opposed to SPIRES method for handling search.
Take a look at the information about the new kind of search here, or just experiment with it yourself. Please let us know your search experiences, either with the SPIRES or the INSPIRE search methods. Hearing about your search experience will enable us to improve it!
INSPIRE runs on the Invenio search engine, making it much faster and more powerful than SPIRES. You have a choice among various search methods.
The easiest is “Google-like” search. There’s probably no quicker way to find a couple of papers by Thomas and Crewther about quarks in 2002 than just typing “thomas crewther quark 2002” into the search box.
Invenio syntax allows smart searches that were impossible with SPIRES. Say you’re interested in finding papers that Parke has written, and that Witten has cited, the author: and the citedby: operators are for you: author:parke citedby:author:witten“. You could alternatively look for the papers that Witten has written which refer to these papers by Parke, with the refersto: operator “author:witten refersto:author:parke“
Of course, SPIRES syntax is there…to stay. Just remember to put “find” in front of your query to tell our search engine you’d like to use SPIRES-style searching. You can also use the refers to and cited by in SPIRES syntax “find date last month and refersto a ellis” to find recent papers referring to papers by Ellis.
We’ve seen from logfiles, and a few of you have contacted us (thanks!), that most people wanted SPIRES-like searching and that many searches failed because of a missing “find”. So, just remember to put “find” in front of your queries.
So why are there several ways to search in INSPIRE? Why not just accept SPIRES-style searches only? Because there are already a large fraction, though not a majority, of users who enjoy Google-like and Invenio syntaxes. Initial feedback is that those are more natural, more powerful, and/or more succinct for their purposes. So our search interface has to balance the different needs of many users, those who love the tried and tested SPIRES-style syntax and those (on the rise) who requested Google-like syntax. We strive to be useful for all.
Driven by user feedback we are working to improve the search experience for our users. We’ve made small changes, like adding a more prominent note about using “find” to trigger SPIRES searching or removing a drop-down menu with a list of search types which caused some confusion. At the same time, we are working on ways to smartly detect what syntax is meant. You might type a few words and we would try to figure out if you meant a SPIRES syntax, an Invenio one, or a Google-style one. This is nontrivial to do without sacrificing some search speed. Since many of the failed searches in our log files were author searches we’ve done a first step in that direction: a query starting with “a ” or “author ” is now interpreted as a search in SPIRES syntax. If our log files show a decrease of search failures in the near future we’ll also include other common SPIRES search terms like “t” for title.
We will continue to improve the search interface of INSPIRE – but we can only do this if we know what’s wrong. If something annoys you, we’ll cannot fix it unless you let us know. We’d much rather hear a complaint from someone who cares enough to send it than silence from someone who doesn’t care enough to help the community improve its tools.