| enhirian on demand | ||||
|---|---|---|---|---|
| Team A | Team Java | Team Itchy | Baseline | |
| ensuring on demand 1.00 | enh irina on demand 0.10 enh irian on demand 0.10 en irian on demand 0.08 | enhirian on demand 1.0 | endhiran on demand 0.9 enhirian on demand 0.1 | |
| This was direct from my search log - I was looking for an on-demand version of the Indian/Tamil movie Endhiran, which is basically the Indian Terminator and is hilarious/awesome. Google finds the right correction, but I don't know if that's because I'm logged in. In retrospect, maybe the results vary due to capitalization? |
||||
| saark alias | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| sock alias 1.0000 | soak alias 0.42 sark alias 0.15 saark alias 0.10 | saark alias 1.0 | sark alias 0.9 saark alias 0.1 | |
| This may have been from my search log - I can't remember how to spell Sark (a bad guy from Alias). I feel like the 'a' is drawn out in saying it, so I always want two a's. Google does fine with this one. Again I wonder if capitalization matters for the project solutions. |
||||
| tom wei villanova | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| tom we villa nova 1.00 | tom we villanova 0.49 tom wei villanova 0.06 tom weir villanova 0.05 | tom wei villanova 1.0 | tom wei villanova 1.0 | |
| I was looking for Tom Way's webpage - he's a UD CIS alum and faculty at Villanova. I couldn't remember how to spell his last name. Interestingly, Google doesn't correct this but the correct result is #6 anyway. Our teams correct "wei" to "we", I'm guessing because they prefer common nouns? |
||||
| britney spear | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| britney spears 1.00 | britney spear 0.56 brittney spear 0.19 britteny spear 0.09 | britney spear 1.0 | britney spears 1.0 | |
| This was an example I saw in a paper - there are many ways of spelling the first name. Unfortunately, I guessed the right one. Google outright rejects my spelling. It doesn't even show "Searching instead for". Team A nails this one, but Team Java doesn't have the correction of "spear" at all. |
||||
| keith trnka | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| keith frank 0.79 keith track 0.21 | keith drink 0.35 keith trina 0.12 keith trnka 0.09 | keith trnka 1.0 | keith trnka 1.0 | |
| Well, one team got it with p=0.9. That's better than nothing, right? | ||||
| kathleen f mccoy | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| kathleen mc 0.45 kathleen major 0.30 kathleen mac 0.25 |
kathleen f mccoy 0.49 cathleen f mccoy 0.08 kathleen f mckay 0.06 | kathleen f mccoy 1.0 | kathleen f mccoy 1.0 | |
| Team Java's gets this right on. The alternate spelling of Kathleen is impressive too. | ||||
| weathr | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| with 0.68 weather 0.16 whether 0.16 | whether 0.53 weather 0.17 weathr 0.09 | weathr 1.0 | weather 0.9 weathr 0.1 | |
| Team A returns some interesting results. You can see what happened - "with" is so common, but we'd never use that in a search. This might be a case of mismatch between query language and normal web language. Also Team Java has "whether" as top, which is better than "with" I think, but neither are content words anyway. Both have "weather" at #2. | ||||
| abbreviatons and acryonyms webpage | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| abbreviations and acronyms web page 0.83 abbreviations and acronyms page 0.17 | abbreviatons and acryonyms web page 0.17 abbreviatons and acronyms web page 0.16 abbreviations and acryonyms web page 0.13 | abbreviatons and acryonyms webpage 1.0 | abbreviatons and acryonyms webpage 0.6 abbreviations and acronyms web page 0.4 | |
| Google shows "Did you mean" instead of "Showing results for". But yet the results look right. Team A corrects it, but splits "webpage" (which is probably fine). Team Java noted that they didn't explore the cross-product of errors really. |
||||
| font used on ligitech illuminated keyboard | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| font used on legit illuminated keyboard 1.00 | font used on logitech illuminated keyboard 0.24 font used on ligitech illuminated keyboard 0.18 font used on ligated illuminated keyboard 0.12 | font used on ligitech illuminated keyboard 1.0 | font used on logitech illuminated keyboard 0.9 font used on ligitech illuminated keyboard 0.1 | |
| This was a legit typo while making the list of typos. Should be Logitech. Team Java gets this at #1, which is impressive for a proper noun that isn't capitalized. It seems like Team A's design may be more biased towards common nouns? |
||||
| univrsity of delawar | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| university of delaware 1.00 | univrsity of delaware 0.23 university of delawar 0.18 univrsity of delawar 0.11 | univrsity of delawar 1.0 | university of delaware 0.9 univrsity of delawar 0.1 |
|
| Team A's results are really impressive on this one. Team Java gets burned due to predominantly single-word corrections. | ||||
| meme guile theme | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| meme guile theme 1.00 | meme guile theme 1.0 | meme guile theme 1.0 | meme guile theme 1.0 | |
| I was searching for a meme called "Guile's theme goes with everything". All of these are fine, but I'd prefer the "'s" attached. |
||||
| french quotation mark | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| french quotation marks 0.6606 french quotation mark 0.3394 | french quotation mark 1.0 | french quotation mark 1.0 | french quotation mark 1.0 | |
| Interesting results - at least the first pass of search engines stems words, so it doesn't have a huge effect. But I agree with Team A's results - "marks" sounds more natural. Team Java doesn't offer suggestions because it only corrects non-words (I think). |
||||
| memes bad dudes | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| mem bad dude 0.72 mem bad dudes 0.28 | memes bad dudes 1.0 | memes bad dudes 1.0 | memes bad dudes 1.0 | |
| I was searching for a meme called "Bad Dudes". It's interesting that the results of Team A are so different with the "s" on "meme", compared to the Guile one. Also, I'm a little surprised that "mem" is in their dictionary. |
||||
| whether 19711 | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| whether 19711 1.00 | whether 19711 1.0 | whether 19711 1.0 | weather 19711 0.9 whether 19711 0.1 | |
| This one was an artificial phonetic real-word error - I thought doing it without the 19711 would be too hard, so I included that. Both listed teams have trouble because it's a real-word error. |
||||
| fonebook | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| feedback 0.69 funk 0.16 funky 0.15 | funk 0.58 fink 0.11 phonebook 0.09 | fonebook 1.0 | fonebook 1.0 | |
| Google leaves this one because there are many sites called "fonebook" or user names. I think Tim said it best - after all, we probably always want funky links. Team Java gets this correct at position #3. |
||||
| c++ trie implementations | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| c++ trie implementation 0.82 c++ trie implementations 0.18 | c true implementations 0.23 c trie implementations 0.23 c tire implementations 0.14 | c++ trie implementations 1.0 | c++ trie implementations 1.0 | |
| This one was added because I figured people might correct "trie" to "tree". The plural/singular of Team A is interesting. Ultimately the "++" is preprocessed away, which the Team Java results show. |
||||
| bananadoor | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| banana door 0.54 banned 0.24 banana 0.21 | banana door 1.0 | bananadoor 1.0 | bananadoor 1.0 | |
| The term "bananadoor" has been used in several artificial word sense disambiguation tasks, including one that Rich Burns did. The Google results here are very interesting - it doesn't correct the query but includes results containing "banana-door" and "banana door". |
||||
| solaris getrusage | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| solar is picturesque 0.43 solar is cartridge 0.31 solar is grotesque 0.25 | is getrusage 0.37 solaris getrusage 0.09 solaris grotesque 0.08 | solaris getrusage 1.0 | solaris getrusage 1.0 | |
| getrusage is "get resource usage" and is the way to measure memory/etc usage in C++ in Solaris and some other *NIX operating systems. Team Java has it correct at #2. In this case, attempting to correct only one thing at a time probably helped. |
||||
| what is .ics | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| what is .ics 1.00 | what is cs 0.46 what is ics 0.18 what is ivs 0.15 | what is .ics 1.0 | what is .ics 1.0 | |
| I was trying to figure out what a particular file extension was. I'm not sure why there's a difference between the two systems. |
||||
| avogadros number | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| avogadro number 1.00 | avocado s number 0.51 avogadro s number 0.18 avogadro number 0.12 | avogadros number 1.0 | avogadro's number 0.9 avogadros number 0.1 | |
| I couldn't remember what Avogadro's constant/number was. Team A gets something good enough (though I'd prefer the "'s" in there). Team Java has that result at #2, but below "avocado's number". |
||||
| create pdfs in java | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| create pets in java 1.00 | create pdfs in java 0.48 create pods in java 0.11 create pads in java 0.11 | create pdfs in java 1.0 | create pdfs in java 1.0 | |
| I was looking for an API to generate PDFs in Java. Team A probably doesn't have "pdfs" in their dictionary. Team Java comes through on this one. |
||||
| is there a way to make justin.tv use html5 | ||||
| Team A | Team Java | Team Itchy | Baseline | |
| is there a way to make justin.tv use html5 0.75 is there a way to make justin.tv uses html5 0.25 | is there a way to make justin tv use html5 0.15 is there a way to make justin tb use html5 0.14 is there a way to make justine tv use html5 0.10 | is there a way to make justin.tv use html5 1.0 | is there a way to make justin.tv use html5 1.0 | |
| Adobe Flash on OS X is terrible, but HTML5 video is fine, so I was trying to see if a particular site could be forced to use HTML5 like YouTube. I'm pretty surprised that both teams had the correct (original) as #1. |
||||