how to avoid an item in an array be repeated in another cluster?

Question

I have an array or objects, approx 58000 unique strings (partName).

parts = [
{ _id: 59a942a8c0b7467bf08711df, partName: '0' },
{ _id: 59a94299c0b7467bf084a917, partName: '9129' },
{ _id: 59a94299c0b7467bf084a918, partName: '9130' },
..,
.. ]

Here is the code. What I'm trying to create clusters of similar strings.

The code does this, but a partName that already is connected with a cluster can appears again in another cluster. I want to avoid this. One unique part name should only be connected to ONE cluster.

Here is an example I get:

{
    "9129": [
      "9132",
      "9190",
      "9279"
    ]
  },
  {
    "9130": [
      "9132",
      "9180",
      "9190",
      "9430"
    ]

As you can see strings 9132, 9190 are repeated in the following cluster. So my question is: How do I remove the strings 9132, 9190 and 9279 after they have been connected with cluster 9129?

function createCluster(arrayOfParts) {
  let clusterArray = [];
  for (var i = 0; i < 5; i++) {
    let cluster = [];
    y = 1;
    console.log(arrayOfParts[i]);
    for (var j = y; j < arrayOfParts.length; j++) {
      if (
        fuzzball.token_sort_ratio(
          arrayOfParts[i].partName,
          arrayOfParts[j].partName
        ) > "70"
      ) {
        if (
          arrayOfParts[i].partName.toLowerCase() !==
            arrayOfParts[j].partName.toLowerCase() &&
          !cluster.includes(arrayOfParts[j].partName)
        ) {
          cluster.push(arrayOfParts[j].partName);
        }
      }
    }
    let obj = {};
    obj[arrayOfParts[i].partName] = cluster.sort();
    clusterArray.push(obj);
  }
  console.log("clusterArray", JSON.stringify(clusterArray, null, 2));
  console.log("clusterArray.length", clusterArray.length);
}

Answer 1

You could define a bank of used strings outside the scope of the loop, add them and check if they have been added to there before adding them again?

Alternatively, if you have a finite (and potentially easily iterable) number of part names, you could hold them as keys with a true/false value to represent 'usable' that you could switch when used.

Solution (using a "bank of used strings" outside the scope of the loop):

function createCluster(arrayOfParts) {
  let usedStrings = [];
  let clusterArray = [];
  for (var i = 0; i < 5; i++) {
    let cluster = [];
    y = 1;
    if (usedStrings.includes(arrayOfParts[i].partName)) {
      continue;
    }
    console.log(arrayOfParts[i]);
    for (var j = y; j < arrayOfParts.length; j++) {
      if (
        fuzzball.token_sort_ratio(
          arrayOfParts[i].partName,
          arrayOfParts[j].partName
        ) > "70"
      ) {
        if (
          arrayOfParts[i].partName.toLowerCase() !==
            arrayOfParts[j].partName.toLowerCase() &&
          !cluster.includes(arrayOfParts[j].partName) &&
          !usedStrings.includes(arrayOfParts[j].partName)
        ) {
          cluster.push(arrayOfParts[j].partName);
          usedStrings.push(arrayOfParts[j].partName);
        }
      }
    }
    let obj = {};
    obj[arrayOfParts[i].partName] = cluster.sort();
    clusterArray.push(obj);
  }
  console.log("clusterArray", JSON.stringify(clusterArray, null, 2));
  console.log("clusterArray.length", clusterArray.length);
}