pith. sign in

USPTO: us-12619834 · published 2026-05-05 · patents

Systems and methods for intent classification in a natural language processing agent

Pith reviewed 2026-05-06 03:53 UTC · model claude-opus-4-7

classification patents
keywords intent classificationcross-lingual NLPmultilingual sentence embeddingsweighted K-nearest-neighborfew-shot learningout-of-domain detectiondual encoderconversational AI
0
0 comments X

The pith

A multilingual sentence encoder paired with a weighted nearest-neighbor decoder classifies user intents from one example per intent and rejects out-of-domain inputs by a distance percentile.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The patent claims a recipe for building a chatbot intent classifier that does not need labeled training data in every language it serves. A frozen multilingual language model produces sentence embeddings for a small set of example utterances tagged by intent, and at inference time the user's utterance is embedded into the same space and matched against those examples with a weighted K-nearest-neighbor decoder. If the nearest examples are far enough away — by a threshold derived from a percentile of inter-class distances seen during training — the system declares the input out-of-domain rather than forcing a label. The pitch to a sympathetic reader is operational: one annotated utterance per intent in one language is enough to deploy across languages, and a single distance cutoff replaces a separately trained OOD detector.

Core claim

The patent describes an intent classifier that needs only a single example utterance per intent and works across many languages without per-language training data. A pretrained multilingual language model encodes both the example utterances and the incoming user utterance into a shared embedding space; a weighted K-nearest-neighbor decoder then assigns the intent of the closest examples, or returns out-of-domain when no example is close enough. The threshold for "close enough" is set from a percentile of inter-class distances measured on the training examples, giving a principled cutoff for the OOD decision.

What carries the argument

A dual-encoder configuration of a pretrained multilingual language model with a weighted K-nearest-neighbor classifier serving as the decoder. The encoder is kept frozen; only the weighted KNN parameters are trained against annotated labels via a classification loss. The OOD decision uses a threshold computed as a percentile of pairwise inter-class distances among training utterance embeddings.

If this is right

  • <parameter name="0">Customer-support and voice-assistant deployments can add a new language by translating or curating a single example per intent
  • without retraining the underlying model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • <parameter name="0">Editorial inference: the quality of the system is essentially the quality of the underlying multilingual encoder
  • gains in open multilingual sentence embeddings should translate directly into gains here without any change to the claimed architecture.

Load-bearing premise

That this specific combination — a frozen multilingual sentence encoder, a weighted nearest-neighbor decoder, and a percentile-of-inter-class-distance cutoff for out-of-domain — is distinct enough from existing sentence-embedding retrieval and nearest-neighbor few-shot classification approaches to count as a new invention rather than an assembly of known parts.

What would settle it

Build a baseline using an off-the-shelf multilingual sentence encoder, a weighted KNN over one example per intent, and an inter-class-distance percentile as the OOD threshold, then evaluate on standard cross-lingual intent benchmarks. If the baseline matches or beats the patent's reported behavior, the claim of a novel system collapses into a description of a known recipe.

Figures

Figures reproduced from USPTO: patent/us-12619834 by Anuprit Kale (Oakland, CA), Gurkirat Singh (Elk Grove, CA), Shashank Harinath (Mountain View, CA), Shilpa Bhagavath (Mountain View, CA), Shubham Mehrotra (Santa Clara, CA), Zachary Alexander (Berkeley, CA).

Sheet 1
Sheet 1. Drawing sheet 1 from US 12619834. view at source ↗
Sheet 2
Sheet 2. Drawing sheet 2 from US 12619834. view at source ↗
Sheet 3
Sheet 3. Drawing sheet 3 from US 12619834. view at source ↗
Sheet 4
Sheet 4. Drawing sheet 4 from US 12619834. view at source ↗
read the original abstract

Embodiments described herein provide a cross-lingual intent classification model that predicts in multiple languages without the need of training data in all the multiple languages. For example, data requirement for training can be reduced to just one utterance per intent label. Specifically, when an utterance is fed to the intent classification model, the model checks whether the utterance is similar to any of the example utterances provided for each intent. If any such utterance(s) are found, the model returns the specified intent, otherwise, it returns out of domain (OOD).

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Axiom & Free-Parameter Ledger

4 free parameters · 2 axioms · 0 invented entities

The patent rests on standard machine-learning ingredients plus one heuristic (the percentile-based OOD threshold). It does not invent new entities; it composes existing ones. The 'free parameters' are the engineering knobs of the pipeline (K, weighting scheme, threshold percentile, choice of encoder), which are tuned to data but disclosed as such.

free parameters (4)
  • K (number of nearest neighbors)
    Standard KNN hyperparameter; not specified in the excerpt.
  • KNN distance weighting scheme
    The 'weighted' qualifier in the claim is not pinned down in the available text; trained via a loss against annotated labels per claim 4.
  • Inter-domain OOD threshold percentile
    Set from a percentile of inter-class training distances per claim 5; specific percentile not given.
  • Choice of pretrained multilingual encoder
    Treated as an external input; not specified in the excerpt.
axioms (2)
  • domain assumption Pretrained multilingual encoder embeddings cluster cross-lingually by meaning well enough for KNN classification.
    Carries the cross-lingual claim.
  • ad hoc to paper Percentile of inter-class training distances is a usable OOD threshold.
    Introduced in claim 5 without derivation; heuristic.

pith-pipeline@v0.9.0 · 17241 in / 5253 out tokens · 151086 ms · 2026-05-06T03:53:20.567721+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.