Double-check your text! To ensure the quality of your text, don’t rely only on one tool. Double-check online!
There are three major approaches to use a spell checker and check text for errors.
- Each word is compared to a word in a given dictionary. A misspelled word is identified easily as long as the dictionary is large enough to contain the word. This is the simplest method and most spell checkers work like this.
- Some spellcheckers also detect common grammatical errors through predefined rules. In this kind of approach, a spell checker service will detect the wrong use of the indefinite article “”, like using an” instead of “a” before a word beginning with a vowel sound. Another example is to find homophones, words that sound more or less the same, but have been used in the wrong manner. An example is ‘write’ and ‘right’. Especially people with dyslexia know about this problem. For this to work, the spellchecker needs to look at several words at the same time to get the context instead of just looking for one word after another.
- The third approach is to create rules based on statistical information. This approach needs a lot of precomputed data from a large text corpus (yes, Wikipedia is too small for that). The collected data is too big to integrate them on a client computer, so this approach is normally not offered by most spell checkers. Usually a rule based spellchecker has a solid base of grammatical rules, but as you may know, there is always an exception of an exception of a rule. Statistical grammar checker may find errors that a rule based spellchecker does not.
Online-spellcheck.com makes use of all three approaches. We have one of the largest set of dictionaries, with thousands of rules and we use a statistical corpus to find even rare errors. And this not only for the English language, we check text for misspelled words and grammatical errors for over 20 languages.
Our online spell checker can do more than most spell checkers
As explained above, spell checking and grammatical improvements of text can be made using three different main approaches. Our online converter uses all of them. Our servers are also quite powerful with lots of RAM to store the large corpus. They are also constantly updated and improvements are applied. There is no need to install software on every device you own to proofread your text. Just open your browser on any device and you are set. And best of all, our spell checker is completely free.
Create your own rules
No one is perfect and we do not claim to find every error in your text. That is just not possible with a machine-only check. If others claim they can do this automatically, it is just not correct. The last resort is always a human (and even this person may fail from time to time). Nevertheless our online spellchecker will help you find most errors and will also make suggestions for grammatical improvements.
We have included a feature where you can tag words that have been marked as being spelled incorrect or grammatically wrong by our system, but are nevertheless correct (e.g. names). If you have registered with us, this will allow you to mark these words and add them to your own personal dictionary. We will use this information for your further checks.
Furthermore, if you click on a word, we provide a dictionary entry with several possible meanings of said word.
Send us documents in any format and we will convert them and fix the errors
You can choose several ways to submit your text:
- You can either copy the text and paste it in a text box on our online spellchecker.
- You can also provide a URL where we can download the document to spell check from the internet. It is also possible to check a website for errors by just entering the domain name.
- It’s also possible to spell check an uploaded file provided in any given format. We support Microsoft Word (DOC/DOCX), Adobe PDF, Text files, RTF files and can even extract text from images using optical character recognition (OCR).
- If you are using services like Google Drive or Dropbox, you can also upload documents from your cloud storage. The same formats as mentioned under (3) are supported as well.
Just give it a try.
Spell checking is important, but we have more
Once we have received and checked your text, we will display the corrected text. During your manual review of the corrected text, you have several options to handle the information we produce. You can revert a change made by us, use our thesaurus to find synonyms or rearrange sentences. We also give some detailed explanation about why a word might be wrong and include examples. When you are satisfied, you can share your text with others by sending them the unique, non guessable URL to your friends, download the file in various formats (like PDF, DOC, TXT or ODT) or send the text as an attachment to yourself.
Who benefits from an online spell checker
To write a text without any errors is difficult, even for experts. Our automated spell checker can help professionals, students, website owners, bloggers and writers to deliver almost error free text. Especially users that write in a language that is not their mother tongue will benefit from our online tool. It will not only help them to correct the text, but they will also get information, why the use of a word is wrong in the given context.
How does a SPELL CHECK works
We see and use spell corrector everyday whenever we are using google or any mobile device. I always wondered how they correct my wrong spelling and recommend me the correct word.
It starts with checking of non-words and real-words.
Non-words do not appear in dictionary. They are the result of typo. The task of correcting such error involves error detection and substituting correct word.
On other hand real-word error are hard to detected as they appear in dictionary and we have to check all the real-words for their correct word.
Correcting non-word errors
Step 1:
Find out the non-word in a sentence by checking each word whether it exists in the dictionary or not.
Suppose we have found that word “tha” as error word.
Step 2:
Since we can guess the correct word will be “the” or the second best word will be “that”.
But what if the word is “acror”. Here we can guess the words like “actor”, “across” etc.
In general we need a set of real words that are similar to error.
Question: How to choose the words similar to error?
Since we know that error is not in dictionary.The basic approach is to use Minimum Edit Distance Algorithm.
Minimum Edit Distance Algorithm:
Input : Error word and dictionary
Output: a list of words with minimum edit distance.
What are edits?
There are 4 basic edit operation that can be considered.
Deletion, Insertion, Substitution and Transposition of two adjacent alphabets.
Note: Edit Distance is the number of the above operation on error word to get a real word.
In general the edit distance is the number of operations (insertions, deletions or substitutions) needed to transform one string in another.
Operations in details.
delete operation:
deleting a letter from error to get real word.
example:
“controal” — -> after deleting “a” becomes “control”
Here Edit Distance = 1
Insert operation:
We had to insert a letter in error to get a real word
example;
“contrl” → after inserting “o” we get “control”
Here Edit Distance = 1
Substitution operation:
It is obvious we have to substitute a letter to get a real word.
example:
“cuntrol” → after substitution of “o” becomes “control”
Here Edit Distance = 1
Transposition operation:
here we have to swap two consecutive letters.
example;
“contorl” → after trans-positioning “ro” we get “control”
Here Edit Distance = 1
NOTE: If any of the above operation occur once then Edit Distance = 1 and if twice on same error then Edit Distance = 2.
We can only use words with Edit Distance = 1 for sake of simplicity as around 70–80% of words have one error.
For more accuracy we can add words with Edit Distance 2 in our model and that is sufficient for the problem.
Step 2 continued:
Find words that are at Edit Distance 1 and 2 from error.
Now we have a set of real words that are similar to error.
Step 3:
Now our goal is to find the most appropriate word for the error. For the purpose we will use Maximum Likelihood Estimation.
Lets assume x is the typo and w be a word from list of similar word(W) and for each w we have to compute the following.
where P(w) is the prior or the probability of w
and P(x|w) is the likelihood of x given w
Now the task is to compute the prior and the likelihood.
Prior:
Form any given corpus we can directly compute the probability of words(P(w))
Likelihood:
It is a bit complicated task to compute P(x|w). Since we have to measure that how likely the non-word occur given a real-word. In other word we can compute the probability of the operation. Example- In case of substitution of letter “m” in place of “n”, we can compute that how often they are substituted and similarly for other operations.
Result:
Since w_estimate = prior * likelihood
Now we can choose the real-word having the best estimate and our Spell Corrector is ready.
Please feel free to comment and ask questions. Stay tunes for more NLP stuff.