mostlikely

Most-Likely Classification Through Bloom-Filtering

Installation

$ npm install mostlikely

About

MostLikely is a library for JavaScript (for use in the Node and Browser environment), providing a space-efficient probabilistic data structure that can be used to test whether an element is a member of a set. False positive matches are possible, but false negatives are not, i.e., a query returns either "possibly in set" (within a configured error rate) or "definitely not in set" (with a 100% accuracy).

Internally, MostLikely is based on a Bloom Filter, which in turn is based on multiple non-cryptographically hash functions and a bit-field. The essential point of Bloom Filters and hence MostLikely is that it is a very space-efficient data-structure for representing a set.

The distinct features of MostLikely (compared to standard Bloom Filter implementations) are:

• optionally supports element removal operation through a counting Bloom filter variant,
• derives its required hash functions from a full set of distinct hash functions (MurmurHash3, Jenkins, CRC32, DJBX33X and FNV),
• supports exporting/importing through Run-Length Encoding (RLE) compression and Base16/Z85 encoding,
• is a standalone/dependency-free implementation and
• supports both Node and Browser run-time environments.

Example

/*  use MostLikely  */
import MostLikely from "mostlikely"

/*  use Pure-UUID and Chai (just for test-driving)  */
import UUID from "pure-uuid"
import Chai from "chai"

/*  configure the expectations  */
const [ itemCount, errorRate, worstErrorRate ] = [ 1000, 0.005, 0.010 ]

/*  create a new set  */
const ml = new MostLikely(itemCount, errorRate)

/*  create 1000 positive test elements and insert them into the set  */
let uuids1 = {}
for (let i = 0; i < itemCount; i++) {
    uuids1[i] = new UUID(1)
    ml.insert(uuids1[i], 16)
}

/*  create 1000 negative test elements (which are NOT inserted into the set)  */
let uuids2 = {}
for (let i = 0; i < itemCount; i++)
    uuids2[i] = new UUID(1)

/*  check our expectations */
let [ err1, err2 ] = [ 0, 0 ]
for (let i = 0; i < itemCount; i++) {
    if (!ml.contains(uuids1[i], 16)) err1++
    if ( ml.contains(uuids2[i], 16)) err2++
}
Chai.expect(err1 / itemCount).to.be.equal(0)
Chai.expect(err2 / itemCount).to.be.most(worstErrorRate)

Application Programming Interface

MostLikely provides the following API:

• MostLikely::new(itemCount?: Number = 1000000, errorRate?: Number = 0.005, mask?: Boolean = true, counter?: Boolean = true): MostLikely:
  Creates a new MostLikely set, intended for an expected number of itemCount items, an expected "false positive" error rate errorRate and internally using a mask-based bit-field and/or a counter-based bit-field.

• MostLikely::export(type?: String = "rle+z85"): Object:
  Export the MostLikely set into a JSON-encodable object. The returned object is of the following type: { bits: Number, hashes: Number, mask?: String, cntr?: String }. The bits is the number of bits the mask bit-field contains (the cntr bit-field contains bits * 4 bits instead). The hashes is the number of distinct hashes which are used. The mask (if present) is the encoded (and optionally compressed) bit-field of bits number of bits. The cntr (if present) is the encoded (and optionally compressed) bit-field of `bits

  • 4` number of bits.

• MostLikely::import(obj: Object, type?: String = "rle+z85"): MostLikely:
  Import the MostLikely set from a (previously exported) JSON-encodable object.

• MostLikely::format(type?: String = "rle+z85"): String:
  Format the MostLikely set into a network transmitable string representation.

• MostLikely::parse(str: String, type?: String = "rle+z85"): MostLikely:
  Parse the MostLikely set from a (previously formatted) network transmitable string representation.

• MostLikely::insert(data: any[], size?: Number = data.length): MostLikely:
  Insert an element into the MostLikely set. The element has to be an Array-like data structure (usually either a real Array, a typed array like Uint8Array or a String).

• MostLikely::remove(data: any[], size?: Number = data.length): MostLikely:
  Remove an element from the MostLikely set. The element has to be an Array-like data structure (usually either a real Array, a typed array like Uint8Array or a String). This operation requires the MostLikely set to be instanciated with counter enabled.

• MostLikely::contains(data: any[], size?: Number = data.length): Boolean:
  Checks whether an element was inserted into the MostLikely set. The element has to be an Array-like data structure (usually either a real Array, a typed array like Uint8Array or a String). If this function returns false, you can be 100% sure the element was not inserted into the set. If this function returns true it is most-likely the element was inserted into the set, but it can be also a false positive (expected to happen only with a probability of the configured errorRate). In other words: on errorRate=0.005 you have to expect that this function returns true in 5% of all cases even for elements which were NOT inserted into the set.

• MostLikely::clear(): MostLikely:
  Removes all previously inserted elements from the MostLikely set.

Implementation Notice

Although MostLikely is written in ECMAScript 6, it is transpiled to ECMAScript 5 and this way runs in really all(!) current (as of 2016) JavaScript environments, of course.

Additionally, there are two transpilation results: first, there is mostlikely.browser.js for Browser environments. This is a size-compressed variant. Second, there is mostlikely.node.js for Node.js/IO.js environments. This is a variant without compression.

License

Copyright (c) 2016-2023 Dr. Ralf S. Engelschall (http://engelschall.com/)

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.