This Arabic version of the PHP similar_text() function provides a robust way to compare two Arabic strings by calculating how closely they match, both character-by-character and in their overall structure. Internally, it utilizes the Needleman-Wunsch algorithm, a well-known sequence alignment technique, which assigns scores to matches, mismatches, and gaps (insertions or deletions). Three key factors influence how characters are scored against each other: keyboard proximity, graphical similarity, and phonetic similarity. Keyboard proximity involves checking how close two characters are on a typical Arabic keyboard layout; graphical similarity considers how closely characters share certain shapes; and phonetic similarity groups characters that produce similar sounds.
By analyzing these three factors, the similar_text() function combines them into a single measure of similarity for each pair of characters, and from there produces a total alignment score for the entire pair of strings. As an additional feature, the function allows retrieval of this score as a raw value and as a percentage. The percentage expresses the comparison result as a fraction of the maximum possible alignment score, giving an intuitive measure of overall closeness between the two strings.
Another powerful aspect of this functionality lies in its configurability. The setSimilarityWeight() method enables you to assign relative importance to keyboard, graphical, or phonetic similarities. If you wish to emphasize one factor (such as phonetic closeness) more strongly than others, you can increase its corresponding weight. Conversely, you can de-emphasize or even ignore one of the factors by reducing its weight to zero. This allows fine-grained control over how much each type of similarity influences the final measure, making the comparison flexible enough to cater to diverse use cases, from spell-checking and autocorrection to natural language processing and search optimization.
Example Output 1: