Find different word between two strings

Question

I have a large list of phrases such as

"Nola jumped off the cliff"
"Loroy jumped off the cliff"
"Nola jumped off the couch"
"Leroy lept off the couch"

I need to find each point in a phrase that is a different word and add that word to a node, which is a list of words that can be used in that position in a phrase. So we would end up with.

"Node1(1) Node2(1) off the Node3(1)"
"Node1(2) Node2(1) off the Node3(1)"
...etc

Where node 1 represents a list of the names(Nola,Leroy), node2 represents a list of the actions(jumped,lept) and node3 ends up representing the list of locations(cliff,couch)

The idea is to take a list of the phrases, and have it automatically create the nodes and fill it with the words that can be used at that node in a phrase.

So, 1st how would I generate the list of phrase nodes? I haven't been able to figure out how to compare two sentences and see if they are exactly alike minus one word.

2nd once I have the nodes set up, what would be the best way to compare all the combinations of the nodes to come up with new matches? (hope that made sense)

Answer 1

Nice one, I like it. Since you tagged your question with C#, I wrote the answer also in C#.

A fast way to get the different words between two phrases:

string phrase1 = "Nola jumped off the cliff";
string phrase2 = "Juri jumped off the coach";

//Split phrases into word arrays
var phrase1Words = phrase1.Split(' ');
var phrase2Words = phrase2.Split(' ');

//Find the intersection of the two arrays (find the matching words)
var wordsInPhrase1and2 = phrase1Words.Intersect(phrase2Words);

//The number of words that differ 
int wordDelta = phrase1Words.Count() - wordsInPhrase1and2.Count();

//Find the differing words
var wordsOnlyInPhrase1 = phrase1Words.Except(wordsInPhrase1and2);
var wordsOnlyInPhrase2 = phrase2Words.Except(wordsInPhrase1and2);

Instead of matching the elements yourself by looping over and checking each element, you can save yourself time and use the built-in LINQ functions Intersect, Except, etc...

For creating phrases by random, please refer to the answer of NominSim.

Answer 2

Yet another Linq-based solution that generates all possible combinations:

var phrases = new List<string> {
           "Nola jumped off the cliff",
           "Loroy jumped off the cliff",
           "Nola jumped off the couch",
           "Leroy lept off the couch"
                           };

var sets = (from p in phrases
            from indexedWord in p.Split(' ').Select((word,idx) => new {idx,word})
            group indexedWord by indexedWord.idx into g
            select g.Select(e => e.word).Distinct()).ToArray();


var allCombos = from w1 in sets[0]
                from w2 in sets[1]
                from w3 in sets[2]
                from w4 in sets[3]
                from w5 in sets[4]
                select String.Format("{0} {1} {2} {3} {4}.", w1, w2, w3, w4, w5);

Doesn't make for the most readable code, but was fun writing. =)