# US 2019/0377784 A1 · Automated nonparametric content analysis for information management and retrieval

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0377784 A1
- Document ID: US20190377784A1
- Status: Published application
- Kind: A1
- Application number: 16/415,065
- Filed: 2017-11-16
- Published: 2019-12-12
- Earliest claimed: 2016-11-22
- Assignees: President and Fellows of Harvard College
- Applicants: President and Fellows of Harvard College
- Inventors: Gary KING; Connor T. JERZAK; Anton STEZHNEV
- Web page: https://patentgrep.com/patent/US20190377784A1

## CPC

- G06N 20/00: Machine learning
- G06F 16/00: Information retrieval; Database structures therefor; File system structures therefor
- G06F 18/2132: based on discrimination criteria, e.g. discriminant analysis
- G06F 18/2135: based on approximation criteria, e.g. principal component analysis
- G06F 18/2155: {characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling}
- G06F 18/241: relating to the classification model, e.g. parametric or non-parametric approaches
- G06F 40/151: Transformation
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06V 30/418: Document matching, e.g. of document images
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Embodiments of the invention utilize a feature-extraction approach and/or a matching approach in combination with a nonparametric approach to estimate the proportion of documents in each of multiple labeled categories with high accuracy. The feature-extraction approach automatically generates continuously valued text features optimized for estimating the category proportions, and the matching approach constructs a matched set that closely resembles a data set that is unobserved based on an observed set, thereby improving the degree to which the distributions of the observed and unobserved sets resemble each other.

## Claims

### Claim 1

A method of computationally estimating a proportion of data elements in a plurality of data categories, the method comprising:
(a) receiving and electronically storing a first set of elements, each element in the first set being computationally assigned to one of the plurality of categories and having one of a plurality of feature profiles computationally associated therewith;
(b) receiving and electronically storing a second set of elements, each element in the second set having one of the feature profiles computationally associated therewith;
(c) computationally defining a continuous feature space having a plurality of numerical variables representing the plurality of feature profiles in the first set, the feature space being configured to discriminate between the categories and the feature profiles;
(d) computationally constructing, based at least in part on the first set, a matched set that substantially resembles the second set, each element in the matched set being associated with a plurality of numerical variables representing a plurality of feature profiles associated therewith; and
(e) estimating a distribution of the elements in the second set over the categories based at least in part on (i) the numerical variables associated with the feature profiles in the matched set, and (ii) the feature profiles associated with the elements in the second set.

### Claim at position 2

2-20. (canceled)

### Claim 21

Apparatus for computationally estimating a proportion of data elements in a plurality of data categories, the apparatus comprising:
a computer memory;
a non-transitory storage device for data storage and retrieval; and
a computer processor configured to:
(a) receive and electronically store a first set of elements in the memory, each element in the first set being computationally assigned to one of the plurality of categories and having one of a plurality of feature profiles computationally associated therewith;
(b) receive and electronically storing a second set of elements in the memory, each element in the second set having one of the feature profiles computationally associated therewith;
(c) computationally define a continuous feature space having a plurality of numerical variables representing the plurality of feature profiles in the first set, the feature space being configured to discriminate between the categories and the feature profiles;
(d) computationally construct, based at least in part on the first set, a matched set that substantially resembles the second set, each element in the matched set being associated with a plurality of numerical variables representing a plurality of feature profiles associated therewith; and
(e) estimate a distribution of the elements in the second set over the categories based at least in part on (i) the numerical variables associated with the feature profiles in the matched set, and (ii) the feature profiles associated with the elements in the second set.

### Claim 22 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to computationally create, with respect to the feature space, an element-feature matrix data structure having rows for at least some of the elements in the first and second sets and columns for the feature profiles associated therewith.

### Claim 23 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to computationally create a projection matrix data structure for projecting the element-feature matrix onto a lower-dimensional subspace matrix data structure, wherein the distribution of the elements in the second set over the categories is estimated based at least in part on the numerical variables in the lower-dimensional subspace matrix data structure.

### Claim 24 (depends on 23)

The apparatus of claim 23, wherein the computer processor is further configured to optimize the projection matrix by maximizing an equally weighted sum of a category-discrimination metric and a feature-discrimination metric.

### Claim 25 (depends on 24)

The apparatus of claim 24, wherein the computer processor is further configured to optimize the projection matrix using the Hooke-Jeeves algorithm.

### Claim 26 (depends on 23)

The apparatus of claim 23, wherein the projection is linear, nonlinear or random.

### Claim 27 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to:
(i) identify, for each element in the second set, three nearest neighbors among the elements from the first set; and
(ii) identify the elements in the first set that are closer than a median nearest neighbor among the three nearest neighbors of all elements in the second set,
wherein the matched set is constructed by pruning out the elements that are not identified in steps (i) and (ii) in the first set.

### Claim 28 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to filter the elements in at least one of the first or second set as so to retain information of interest only.

### Claim 29 (depends on 21)

The apparatus of claim 21, wherein the elements in the first and second sets comprise text and the computer processor is further configured to convert the text to lowercase and removing punctuation marks.

### Claim 30 (depends on 21)

The apparatus of claim 21, wherein the elements in the first and second sets comprise text and the computer processor is further configured to map a word to its stem.

### Claim 31 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to summarize the feature profiles in at least one of the first or second set as a set of dichotomous variables.

### Claim 32 (depends on 21)

The apparatus of claim 21, wherein the distribution of the elements in the second set over the categories is not constrained to be the same as a distribution of the elements in the first set over the categories.

### Claim 33 (depends on 21)

The apparatus of claim 21, wherein the distribution of the elements in the second set over the categories is unbiased.

### Claim 34 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to store the distribution of the elements in the second set over the categories on at least one of the computer memory or the storage device.

### Claim 35 (depends on 21)

The apparatus of claim 21, wherein the elements in the first and second sets comprise at least one of text, audio, or video data encapsulated in at least one of files, streams, or database entries.

### Claim 36 (depends on 35)

The apparatus of claim 35, wherein the feature profiles indicate whether certain words occur in the text.

### Claim 37 (depends on 35)

The apparatus of claim 35, wherein the feature profiles indicate whether certain combinations of words occur in the text.

### Claim 38 (depends on 35)

The apparatus of claim 35, wherein the text is unstructured.

### Claim 39 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to analyze at least some of the elements in the first set or the second set to obtain the feature profiles associated with said elements.

### Claim 40 (depends on 21)

The apparatus of claim 21, wherein the computer processor is further configured to estimate the distribution of the elements in the second set over the categories without assigning the elements in the second set to the categories individually.