0

What percentage of words or queries are misspelled in search queries?

I couldn't find any decently recent study. {3} states:

Dalianis measured that 10% of web search engine queries were misspelled {1}. Wang et al. counted as misspellings 26% of the total of unique query terms {2}. We analyzed a random sample of 1 000 queries of the Portuguese Web Archive (PWA) and detected that 5% were misspelled.

but {3} was published 7 years ago, used as a small corpus, I would be interested in having results for English as well ({1} is on Swedish) and {1,2} were published almost 20 years ago.

{4} also used references that are ~20 year-old:

A number of studies of search engine queries have observed a high misspelling rate (Nordlie 1999; Spink, Wolfram et al. 2001; Wang, Berry et al. 2003). Wang, et al. (2003) report a misspelling rate of 26% for words on an academic site.

Note that the percentage of words are misspelled in search queries is an lower bound of the percentage of misspelled search queries. I am more interested in the percentage of words are misspelled in search queries so that one doesn't have to consider the lengths of the search queries.

References:

I have crossposted the question at:

Improve this question
13
  • 4
    My bet is on an increase of misspellings: Since search engines generously correct wrong spellings no one cares to look twice on SE input.
    – Sir Cornflakes
    Apr 9, 2020 at 12:10
  • @jk-ReinstateMonica good point, true, at the same time one could argue 1) increased use of speech recognition and swipe 2) better automated client-side spell checking 3) better typing ability.
    – Franck Dernoncourt
    Apr 9, 2020 at 12:11
  • I assume you mean universal misspellings and not dialect-relative misspellings. Colour is a mispelling in my dialect.
    – user6726
    Apr 9, 2020 at 16:06
  • @user6726 yes, universal misspellings.
    – Franck Dernoncourt
    Apr 9, 2020 at 16:07
  • I bet on an increase too, because of the user expectation that it will be handled, and because of the shift to mobile and generally less literate users who may not be searching in their mother tongue.
    – Adam Bittlingmayer
    Apr 9, 2020 at 21:26

1 Answer 1

Reset to default
2

I found this more recent paper: Duan, Huizhong, and Bo-June Hsu. "Online spelling correction for query completion." In Proceedings of the 20th international conference on World wide web, pp. 117-126. 2011. https://dl.acm.org/doi/abs/10.1145/1963405.1963425:

Our dataset for training the transformation model contains 1.4 million recourse link clicks. The statistics of the training data are shown in Table 2. Around 80% of all queries and 70% of all unique queries are correctly spelled. 1/10 of the training data is held out for parameter tuning.

It lists the following types of misspellings, which is quite keyboard-oriented:

enter image description here

which nicely complements https://www.dailywritingtips.com/7-types-of-misspellings/'s list:

  • Incorrectly repeated consonants
  • Wrong vowel
  • Wrong consonant
  • Reversed order of double vowels
  • Extra letters
  • Missing letters
  • Confusion with similar word
Improve this answer
2
  • 1
    Can we just assume that exxit is a misspelling of exit or silver light of silverlight? Both exist. It's hard to know without analysing the session intent or asking the user. And the other way around could happen too - a misspelling of the intended word that happens to be the correct spelling of another word.
    – Adam Bittlingmayer
    Apr 10, 2020 at 6:50
  • 1
    @AdamBittlingmayer agreed, sometime classifying the misspelling isn't 100% clear without further analysis
    – Franck Dernoncourt
    Apr 10, 2020 at 8:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.