Cordial

History

Conceived by the Toulouse company Synapse Development, Cordial is the fruit of the effort of a team of ten developers and linguists. Born from a correct orthographical integrated in 1988 in the data-processing software the Writer (on Atari ST), which moulted in correct grammatical in 1992 in Writer 4, Cordial knew his first version under Windows in February 1995.
In fact, the “Cordial” word is an acronym for “Corrector Of Inaccuracies and Lexico-syntactic Analyseur”. Since 1995, Cordial knows approximately a new version per annum. The last version, marketed in October 2007, is the version 14 but it is in fact fifteenth because there was a version 3.5 in 1997.
Since 1997, the correct orthographical, the dictionary of Synonymous and the tool with Césure provided with the continuation Office from Microsoft are tools developed consequently company, Synapse Development. On Macintosh, all these tools and the grammatical Correcteur for French results also from this company.

Cordial software

The analyzer

To correct a text automatically, it is necessary grammatically to analyze each sentence. This analysis cannot be carried out as a analyzes grammatical logical usual. Indeed, to locate the faults, the analyzer is constrained to use fuzzy rules of analysis . Thus, in the sentence “they are large consumers of pastry makings”, a formal analyzer will never set up a group between the words “large” and “consuming”, because they are different kinds, therefore either it will connect “large “to “pastry makings”, or it will leave “large like a group with share except grammatical constraint. A fuzzy analyzer, as that uses Cordial, fascinating of account the position of the words in the sentence and the regularities statistical, will place in the same “large” group the adjective and the noun “consumers”, which enables him to detect the fault of agreement and to propose to replace “large” by “large”.
The analyzer of Cordial can be observed in the software, which proposes the analysis in list or tree of each sentence. One of its most spectacular characteristics is its speed since it analyzes a whole work in a few seconds (the authors announce a speed of analysis of 15.000 words/second on Pentium 3 GHz).

The corrector

The corrector applies the grammatical rules to the analysis carried out by the analyzer. When a word is unknown dictionaries (i.e. generally when it is faulty because the dictionaries of Cordial are very complete), it proposes one or more words of replacement, generally only one and almost always the good. It is what is called in data processing the orthographical correction. When a word is in the dictionaries but corresponds to a fault of agreement or, more generally, with a grammatical mistake, the corrector proposes to replace it by the good mot.
The corrector corrects very many types of faults and, in particular, the faults of homophony and homography. Thus, if you write “the page asks for jam of corners. ”, Cordial will propose “the page to you asks for quince jam. ”, while locating the fault of homography enters “the page” and “the page” and the fault of homonymy between “corner” and “quince”.
This level of correction is obtained thanks to the semantic analyzer associated with the grammatical analyzer. The software indeed locates the direction of each word which can have several directions (the found directions are presented in the analysis of the sentence suggested by Cordial).
The orthographical correction is also proposed for English, Spanish and German. In the professional version, the grammatical Portuguese and Italian corrections are added to it.

Performances of the correction

Cordial authorizes very many parameter settings of the correction. Since his first version, one can regulate the level of correction according to the desired level. Either the software finds the vast majority of the errors but at the price of some nonrelevant messages, or it corrects a little less errors but with an excellent probability.
The analyzes grammatical and the location of the directions of the words is not however exact sciences. They are fields of research and the correction of 100% of the errors, especially without any inadequate message, is undoubtedly not for tomorrow. Even certain agreements of the last participle are still the debate object, in particular on the fixed reflexive verbs like “being done extremely”.

Lexical bases

Cordial is characterized by quality and volume from its dictionaries. In the most recent version (Cordial 2008), the dictionary of common nouns just: 211 530 lemmas correspondent with: 1107876 words, and the dictionary of proper names contains: 186108 entries. The corrector is also pressed on dictionaries of phrases containing more: 100000 expressions, on a dictionary of initials of: 40584 terms and on a dictionary of: 5206 abbreviations. All these words, expressions, initials and abbreviations have their definition. Cordial giving access to all its dictionaries, the exactitude of these figures can be checked by any user of the software. They are, according to the editor, the greatest existing lexical bases for French.

Other dictionaries

Cordial is accompanied by very many dictionaries: 18 in its standard version, 67 in its professional version. The dictionary of synonymous and antonyms comprises more: 50000 entries for a total of 4 million synonyms. In its professional version, Cordial gives access to the dictionary Littré like to the Trésor of the French language. The two dictionaries were entirely re-examined for a simplified reading. Thus all the abbreviations, very many in the two works, were wide. miters allow, in the Littré, to choose desired information. Dictionaries of false-friends and word for word translation or expression for expression are also available in the professional version, for any language pair between French, English, German, Spanish, Italian and Portuguese.

Basic corpus

The basic corpus is a whole of works and texts used by Cordial for its development (location of the usual faults, the words used and their frequency, etc) and for functions of comparison between these texts and those of the user.
In its first versions, Cordial used “small” a Corpus of: 2600 works and of texts of newspapers, representing approximately 500 million words. In its most recent version, this corpus was carried to: 5000 works and more than 1,2 billion words. By comparison with the Corpora used in linguistic data processing, this corpus is characterized by its size but also by the big part from recent and very recent works (several hundreds of works are after 2000).

Analyzes texts

Beyond the correction, thanks to the dictionaries and in the corpus, Cordial proposes very many options of analysis of text: seek key words, of concepts - key, of sentences - key, statistical analysis, semantic Analyze, analyzes stylistic, co-occurences, segments repeated. It is also possible to compare each one of its texts with: 5000 works of the Corpus basic describes above like with main categories of text (journalistic, commercial, legal, technical).

Integration in other software

Cordial with the characteristic to be able to replace the corrector of Microsoft Office. Thus the underlinings in red (orthography) and green (grammar) carried out during striking in Word those of Cordial, are once installed the software. That is extremely convivial. In addition the many functions of Cordial are available in the ribbon of MS Word, MS Excel, ms PowerPoint and MS Outlook.
Cordial is also integrated in very many software under Windows like the messaging software, Internet Explorer, the software of page layout, etc

Assistances and handbooks

The software is provided with two handbooks to the format pdf representing more than 600 pages. Very many pages of assistance are also proposed for the whole of the functionalities.
A handbook of Grammar of more than: 1100 articles is also provided. It is interesting to raise here that Synapse Development has gracefully put this handbook on line on the Web for more than ten years.

Customers

Cordial addresses to the whole of the French-speaking . Besides it has very many words of the Canadian vocabularies, Belgian, Switzerland, African and West-Indian, and several parameters make it possible to take into account or to announce the terms resulting from these vocabularies.
The customers of the professional version seem to be especially “professionals of the language” (journalists, writers, lawyers, correctors, translators…) while the standard version touches an audience much larger, in particular the teachers, students and high-school pupils.
A specific version of Cordial, Cordial Analyzer, addresses to the laboratories automatic treatment of the natural language. It is marketed at low-costs and it is very much used by the researchers in Linguistique data processing and letters.

Singular uses

Cordial is however a very rich product, so much so that some of its users do not even make use of the correction part, most important. Thus translators of French towards another language developed a formula calculation of estimate starting from the statistical analysis of Cordial, the degree of Lisibilité and grammatical complexity of the sentences. Editors and cabinets of translation use it to compare the C-Ws communication proper names, via the option “occurrences” and to check that, for a work, same the Graphie is always used for the same word (for example Mao Zedong or Kadhafi). The teachers use much the functions of analyzes grammatical and of analysis of text, but also the dictionaries of synonymous definitions and .
In its version 2008, Cordial can also be used to solve anagrams, cross Mots or many sets of letters.

Competitors

Correct 101, of Machina Sapiens
Antidote, of data-processing Druid
Pro Lexis, of Data-processing Diagonal

Distinctions

Cordial received very many prices in the category “produces year”, on behalf of the reviews “Windows News”, Expert PC, PC Purchase, etc.
The person in charge of the development, Dominique Laurent, was elected “French personality of the year” in 2001, by a European jury data-processing journalists. Dominique Laurent is always responsible for the development of the company, it is also member of the Standing Committee of the TALN (Automatic Treatment of the Natural Languages).

Random links: