Why do translators need a better search engine than EUR-Lex?
EUR-Lex contains a great deal of useful information, but its search engine is not adapted to the needs of translators. If a user wants to obtain a translation of a particular term, the whole procedure consists of four steps:
The user enters the search query, selects the language and clicks the search button.
The program searches the database and lists the first 10 hits (or up to 10 hits if there are fewer). By clicking the title of the document, the user obtains the text in the source language.
By clicking one of the languages listed above the document, a new window opens, showing aligned text in source and target languages.
The user enters the search query in the web browser (usually by pressing Ctrl F) and the first hit is shown in context. Other hits can be seen by pressing the F3 button.
The whole process is rather time consuming (three searches are made in EUR-Lex database and one search in the document on the screen) and thus rather user-unfriendly.
EUR+Lex simplifies this process as much as possible: the user defines the search parameters and clicks the search button. On the output page, only the relevant parts of the documents will be shown (the output is similar to a display from a bilingual corpus). At the top of each data block, the Celex number and the title of the original document is specified, and by clicking the links a full text of the document can be seen in source, target or both languages.
The EUR+Lex program performs all of the above-mentioned search steps, so the user sometimes has an impression that the search is slow, because often one has to wait several seconds to get the results. However, by measuring the standard EUR-Lex search time and comparing it with new EUR+Lex search time, it can be proven that the latter is always faster, because user interventions are not needed.
Enter the search term in the entry field, select the source and target languages, and number of hits per document. Search time partially depends on the number of hits, because the program has to search several pages to obtain more results, so the initial setting is relatively low.
Fast search during translationIntelliWebSearch (IWS) is a free program that allows shortcuts to be made to frequently used search sites (e.g. Google, One-look dictionaries – as well as EUR+Lex). IWS also allows you to search local resources (dictionaries on your PC). After installing IWS (this is a very simple operation), it is necessary to enter the following strings in the program settings (IWS. ver. 3):
When a user gets the message that there were no hits, (s)he should first check whether the search term has been written correctly and whether the proper source language has been selected.
Sometimes there is no corpus output, although the title of the document is listed. In spite of extensive research, I must admit that do not know why this happens. In such cases it sometimes helps if you repeat the search. Otherwise you should check the bilingual (aligned) text.
It is interesting that the results (sometimes) differ when using different browsers: the maximum number of concordance hits is obtained when using the Pale moon web browser, slightly lower number of hits is listed in MS Internet Explorer and the least number of hits is shown in Mozilla Firefox and Google Chrome. I have no explanation for this behaviour. (I found Pale moon by chance and it is not my intention to advertise this (free) software.)
The opposite problem occurs when the header is shown, but there are no hits in the document itself. There are two possibilities why this happens:
there is a slight difference between the search term and the term in the document (e.g. the search term is anti lock brake, while the term is written as anti-lock brake in the document)
some key words are added to each document in the EUR-Lex database and these key words are also used during search (e.g. if the search query is mobile, then one of the documents listed is 32015D1477. If you check this document, you cannot find the word mobile in it, but you can find mobility).
Documents that were created before 2005 are often mis-aligned in the bilingual view. EUR+Lex "solves" this problem in such a way that the results are sorted by dates: the newest hits (usually properly aligned) are listed first. If all corpus results are mis-aligned, check the bilingual document(s).