What Is Proximity Case Search?

Explainer · Case Search

Updated for 2026 · Claw Editorial

When you search for a judgment by a party name, even a small spelling difference can mean the search returns nothing, and you miss the case entirely. Proximity case search is the technique that fixes this: it finds judgments where the name you typed is close to, but not identical to, the name in the record. This explainer covers what proximity search is, why name mismatches are especially common in Indian legal databases, and what to look for in a tool that handles them well.

01The problem: why exact search breaks on names

Every legal search tool stores judgment text in a database. When you type a name, the simplest approach is to match it exactly, character by character. This works well for standard English words. It fails badly for names.

Names get recorded differently each time

A party named "Ramachandran" might appear as "Ramchandra", "Ram Chandra", or "Ramachandaran" across different filings, different clerks, and different courts. Each variation is a different string to an exact-match engine. A search for one version will not return the others, even though all three records are about the same person.

Transliteration adds more variation

Indian names are originally in regional scripts, Hindi, Tamil, Telugu, Kannada, and others. When clerks transliterate them into English for court records, they make judgment calls. "Subramaniam" and "Subramaniyam" are the same name written two different ways. There is no single right spelling.

Typing errors happen

Lawyers, clerks, and litigants type under pressure. A single dropped letter or transposed character ("Mehta" versus "Mehtta") is enough to break an exact search. You may never know you missed a judgment because the search returned no error, just an empty result.

02What proximity search is

Proximity search, in the context of case law, means searching for records where a term is near enough to your query, not just exactly equal to it.

The word "proximity" is used in two related but distinct ways in legal search tools, and it is worth knowing both.

Sense 1: name proximity (fuzzy name matching)

In name proximity search, the engine accepts a small number of differences between the typed name and the stored name. "Ramchandra" and "Ramachandran" are close in character-level distance, so the engine treats them as likely the same name. This is the sense most relevant to finding Indian judgments by party name or judge name.

The technical term for this is fuzzy matching or edit-distance matching. The engine counts how many character insertions, deletions, or substitutions are needed to turn one string into the other. If that count is below a threshold, it treats the strings as matching.

Sense 2: term proximity (terms near each other in text)

Some search engines also support proximity in the positional sense: finding documents where two or more search terms appear within a set number of words of each other. For example, a proximity search for "specific performance" AND "injunction" within 10 words would only return judgments where those two concepts appear closely together, filtering out cases where the judgment discusses them in entirely separate contexts.

Proximity search is what separates a search engine that understands names from one that punishes you for a single wrong letter.

03Why proximity search matters especially in Indian courts

Indian case law databases face name-variation problems at a scale that most other legal systems do not, for reasons that are structural.

25 High Courts, each with its own records

A judgment database covering all 25 High Courts plus the Supreme Court pulls records from many independent sources. Transcription conventions vary by court and by era. A name recorded in 1985 may look different from the same name recorded in 2015, even in the same court.

No single national naming standard

India has no single official romanisation standard for personal names. The same Kannada or Malayalam name can be transcribed several ways, all of them defensible. This is not a data quality failure: it is a structural feature of a multilingual country with a large body of historical records.

Historical records predate computers

Judgments from the 1950s through the 1980s were originally typed or handwritten. Digitising them meant another round of transcription. Each round introduced its own variation. A search tool that does not account for this will simply miss older authorities.

Party names, judge names, and advocate names all vary

Proximity search matters for all three. Searching by a judge’s name to find all their rulings on a legal point, or by an advocate’s name to trace their matters, hits the same problem. An exact-match engine forces you to know every variation in advance, which defeats the purpose of search.

04How proximity search works under the hood

You do not need to understand the mathematics to use a proximity-aware search tool well, but knowing the basics helps you judge whether a tool is genuinely doing it.

Edit distance (Levenshtein distance)

The most common approach is edit distance, named after the mathematician who described it. The engine calculates the minimum number of single-character edits (insert, delete, or substitute) needed to turn the query string into the stored string. "Mehta" to "Mehtta" is one insertion, so the edit distance is 1. Most tools set a threshold of 1 or 2 to decide what counts as a match.

n-gram indexing

Another approach breaks every name into overlapping character groups (n-grams). "Ramachandra" broken into 3-grams becomes "Ram", "ama", "mac", "ach", "cha", "han", "and", "ndr", "dra". Two names that share most of their n-grams are treated as similar. This is fast and works well for partial matches.

Phonetic encoding

Phonetic algorithms convert a name to a code based on how it sounds. Names that sound alike get the same code, even if spelled differently. This handles the case where two clerks spelled the same name according to different conventions. For more detail on this specific technique, see the explainer on phonetic name matching.

How they are combined in practice

A well-built case search engine does not rely on just one technique. It runs the query through multiple methods and merges the results, ranked by how confident it is. The result is a list that starts with exact matches and shades into near-matches, so you see the best hits first without manually trying every spelling variant.

05Proximity search vs keyword search vs semantic search

Case search tools use several different techniques, and each solves a different part of the problem. Understanding the difference helps you know what a tool is actually doing when you type a query.

The best modern tools layer all of these. An AI-based search engine will handle the legal meaning of a query (semantic), and simultaneously apply fuzzy or phonetic matching on names within the results, so you are not forced to choose between finding a concept and finding a party.

06What to look for in a search tool that handles names well

When evaluating a case search tool for proximity and name-matching capability, ask these questions.

Does it show near-miss results or only exact matches?

The easiest test: search for a common Indian name with one letter changed. If the tool returns nothing, it is doing exact matching only. A proximity-aware tool will return results and may even flag which names it treated as equivalent.

Does it cover the courts and years you need?

Proximity matching on a shallow database still leaves gaps. If the tool does not hold judgments from the court or time period you need, no amount of fuzzy matching will find them. Check coverage first.

Does it return verified citations?

A tool that finds a case via proximity matching must still give you a citation you can verify and use in court. If it cannot show you the source or the citation cannot be traced back to an official record, it is not safe to rely on.

Is semantic search also present?

Proximity matching handles name variation. Semantic search handles the legal-concept side of the query. You need both. A tool that only does fuzzy name matching will still miss cases where the point of law is the search target, not a party name.

How do the results rank?

Proximity search can generate many near-matches. The ranking of those results matters. The tool should surface the most relevant and authoritative judgments first, not bury them behind marginal hits.

For a broader look at what separates good case search from weak case search, see the guide to the best case search tools in India.

07Where Claw fits

Claw is an all-in-one legaltech platform for Indian advocates, law firms, and corporate legal teams, combining AI-based case search, an AI legal assistant (Legal GPT), case management, and compliance automation across all Indian courts and tribunals. It is positioned as India’s first all-in-one legaltech platform of this kind.

On case search, Claw combines semantic and AI-based search with name-tolerant (proximity and phonetic) matching across 30 crore judgements spanning 25 High Courts from 1980 to 2026 and the Supreme Court from 1950 to 2026. The intention is that a search finds the right judgment even when the party name has been recorded differently across filings or courts. Results come back with verified, court-ready citations in under 5 seconds, so the output is safe to cite, not just a list of fragments to manually verify.

For teams evaluating tools on this dimension, the question to ask any vendor is not just "do you support fuzzy search" but "show me a near-miss name search across high-court records and tell me how the citations are verified". That is the practical test that separates real proximity matching from a marketing claim.

08Frequently asked questions