# US 2019/0318059 A1 · Semantic pattern extraction from continuous itemsets

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

- Number: US 2019/0318059 A1
- Document ID: US20190318059A1
- Status: Published application
- Kind: A1
- Application number: 16/457,209
- Filed: 2019-06-28
- Published: 2019-10-17
- Earliest claimed: 2019-06-28
- Assignees: Intel Corporation
- Inventors: Bikram Baidya; Vivek K. Singh; Allan Gu; Abde Ali Hunaid Kagalwalla; Saumyadip Mukhopadhyay; Kumara Sastry; Daniel L. Stahlke; Kritika Upreti
- Web page: https://patentgrep.com/patent/US20190318059A1

## CPC

- G03F 1/70: Adapting basic layout or design of masks to lithographic process requirements, e.g., second iteration correction of mask patterns for imaging
- G06F 30/398: Design verification or optimisation, e.g. using design rule check [DRC], layout versus schematics [LVS] or finite element methods [FEM] (optical proximity correction [OPC] design processes G03F1/36)
- G03F 1/36: Masks having proximity correction features; Preparation thereof, e.g. optical proximity correction [OPC] design processes
- G06F 16/24578: {using ranking}
- G06F 16/2462: {Approximate or statistical queries}
- G06F 16/2465: {Query processing support for facilitating data mining operations in structured databases}
- G06F 16/248: Presentation of query results
- G06F 2216/03: Data mining
- G03F: PHOTOMECHANICAL PRODUCTION OF TEXTURED OR PATTERNED SURFACES, e.g. FOR PRINTING, FOR PROCESSING OF SEMICONDUCTOR DEVICES; MATERIALS THEREFOR; ORIGINALS THEREFOR; APPARATUS SPECIALLY ADAPTED THEREFOR; (phototypographic composing devices B41B; photosensitive materials or processes for photographic purposes G03C; electrophotography, sensitive layers or processes therefor G03G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A semantic pattern extraction system can distill tremendous amounts of silicon wafer manufacturing data to generate a small set of simple sentences (semantic patterns) describing physical design geometries that may explain manufacturing defects. The system can analyze many SEM images for manufacturing defects in areas of interest on a wafer. A tagged continuous itemset is generated from the images, with items comprising physical design feature values corresponding to the areas of interest and tagged with the presence or absence of a manufacturing defect. Entropy-based discretization converts the continuous itemset into a discretized one. Frequent set mining identifies a set of candidate semantic patterns from the discretized itemset. Candidate semantic patterns are reduced using reduction techniques and are scored. A ranked list of final semantic patterns is presented to a user. The final semantic patterns can be used to improve a manufacturing process.

## Claims

### Claim 1

A semantic pattern extraction method comprising:
performing an entropy-based discretization on a continuous itemset to generate a discretized itemset, individual of the items in the continuous itemset comprising a physical design feature value for one or more physical design features and an event value indicating the presence or absence of one or more manufacturing defects;
identifying a set of candidate semantic patterns from the discretized itemset by performing frequent set mining on the discretized itemset;
reducing the set of candidate semantic patterns to one or more final semantic patterns;
determining a score for individual of the final semantic patterns; and
displaying the final semantic patterns on a display in a ranked order, the ranked order based at least in part on the score for the individual final semantic patterns.

### Claim 2 (depends on 1)

The method of claim 1, wherein the discretized itemset comprises the items of the continuous itemset with one or more of the physical design feature values replaced by a discretized physical design value or range from a set of discretized physical design feature values or ranges for the respective physical design feature.

### Claim 3 (depends on 1)

The method of claim 1, the determining the score for individual of the final semantic patterns comprising determining one or more of an accuracy score, a coverage score, and an interpretability score for the individual final semantic patterns, the score for the individual of the final semantic patterns based at least in part on one or more of the accuracy score, the coverage score, and the interpretability score.

### Claim 4 (depends on 3)

The method of claim 3, wherein the interpretability score for the individual final semantic patterns is based at least in part on the number of physical design features represented by the one or more physical design feature values in the individual final semantic patterns.

### Claim 5 (depends on 3)

The method of claim 3, the reducing the set of candidate final semantic patterns to one or more final semantic patterns comprising determining an independence ranking for individual of the final semantic patterns, the score for the individual final semantic patterns being further based at least in part on the independence ranking of the individual final semantic patterns.

### Claim 6 (depends on 1)

The method of claim 1, further comprising receiving a user selection of at least one of the final semantic patterns and modifying a design rule database based at least in part on the at least one selected final semantic patterns.

### Claim 7 (depends on 1)

The method of claim 1, further comprising receiving a user selection of at least one of the final semantic patterns and modifying an OPC recipe based at least in part on the at least one selected sematic patterns.

### Claim 8 (depends on 7)

The method of claim 7, further comprising generating mask data based at least in part on the modified OPC recipe.

### Claim 9 (depends on 8)

The method of claim 8, further comprising generating one or more photolithography masks based at least in part on the mask data.

### Claim 10 (depends on 1)

The method of claim 1, wherein the reducing comprises representing the candidate semantic patterns in a graph, wherein nodes in the graph are associated with individual of the candidate semantic patterns and edges in the graph connect nodes associated with candidate semantic patterns that are not independent from each other.

### Claim 11 (depends on 10)

The method of claim 10, the reducing further comprising:
performing a greedy algorithm to determine one or more least-connected nodes, the one or more least-connected nodes having zero edges or the least number of edges to neighboring nodes in the graph among the nodes in the graph;
for individual of the one or more least-connected nodes:
collectively identifying the candidate semantic patterns associated with the individual least-connected nodes and the candidate semantic patterns associated with any neighboring nodes connected to the individual least-connected node in the graph as one of the final semantic patterns;
removing the individual least-connected node and any neighboring nodes connected to the individual least-connected node from the graph; and
assigning an independence ranking to the final semantic pattern; and
iteratively performing the greedy algorithm until the graph is empty, the independence rankings of the final semantic patterns identified in earlier iterations of the greedy algorithm being greater than independence ranking of final semantic patterns identified in later iterations of the greedy algorithm.

### Claim 12

A semantic pattern extraction system comprising:
one or more processors;
one or more computer-readable storage media storing instructions thereon for causing the one or more processors to perform a method comprising:
performing an entropy-based discretization on a continuous itemset to generate a discretized itemset, individual of the items in the continuous itemset comprising a physical design feature value for one or more physical design features and an event value indicating the presence or absence of one or more manufacturing defects;
identifying a set of candidate semantic patterns from the discretized itemset by performing frequent set mining on the discretized itemset;
reducing the set of candidate semantic patterns to one or more final semantic patterns;
determining a score for individual of the final semantic patterns; and
displaying the final semantic patterns on a display in a ranked order, the ranked order based at least in part on the score for the individual final semantic patterns.

### Claim 13 (depends on 12)

The semantic pattern extraction system of claim 12, wherein the discretized itemset comprises the items of the continuous itemset with one or more of the physical design feature values replaced by a discretized physical design value or range from a set of discretized physical design feature values or ranges for the respective physical design feature.

### Claim 14 (depends on 12)

The semantic pattern extraction system of claim 12, the determining the score for individual of the final semantic patterns comprising determining one or more of an accuracy score, a coverage score, and an interpretability score for the individual final semantic patterns, the score for the individual of the final semantic patterns based at least in part on one or more of the accuracy score, the coverage score, and the interpretability score.

### Claim 15 (depends on 14)

The semantic pattern extraction system of claim 14, the reducing the set of candidate final semantic patterns to one or more final semantic patterns comprising determining an independence ranking for individual of the final semantic patterns, the score for the individual final semantic patterns being further based at least in part on the independence ranking of the individual final semantic patterns.

### Claim 16

One or more computer-readable storage media storing instructions thereon for causing a computing device to perform a method comprising:
performing an entropy-based discretization on a continuous itemset to generate a discretized itemset, individual of the items in the continuous itemset comprising a physical design feature value for one or more physical design features and an event value indicating the presence or absence of one or more manufacturing defects;
identifying a set of candidate semantic patterns from the discretized itemset by performing frequent set mining on the discretized itemset;
reducing the set of candidate semantic patterns to one or more final semantic patterns;
determining a score for individual of the final semantic patterns; and
displaying the final semantic patterns on a display in a ranked order, the ranked order based at least in part on the score for the individual final semantic patterns.

### Claim 17 (depends on 16)

The one or more computer-readable storage media of claim 16, wherein the discretized itemset comprises the items of the continuous itemset with one or more of the physical design feature values replaced by a discretized physical design value or range from a set of discretized physical design feature values or ranges for the respective physical design feature.

### Claim 18 (depends on 16)

The one or more computer-readable storage media of claim 16, the determining the score for individual of the final semantic patterns comprising determining one or more of an accuracy score, a coverage score, and an interpretability score for the individual final semantic patterns, the score for the individual of the final semantic patterns based at least in part on one or more of the accuracy score, the coverage score, and the interpretability score.

### Claim 19 (depends on 18)

The one or more computer-readable storage media of claim 18, wherein the interpretability score for the individual final semantic patterns is based at least in part on the number of physical design features represented by the one or more physical design feature values in the individual final semantic patterns.

### Claim 20 (depends on 18)

The one or more computer-readable storage media of claim 18, the reducing the set of candidate final semantic patterns to one or more final semantic patterns comprising determining an independence ranking for individual of the final semantic patterns, the score for the individual final semantic patterns being further based at least in part on the independence ranking of the individual final semantic patterns.

### Claim 21 (depends on 16)

The one or more computer-readable storage media of claim 16, the reducing comprising:
representing the candidate semantic patterns in a graph, wherein nodes in the graph are associated with individual of the candidate semantic patterns and edges in the graph connect nodes associated with candidate semantic patterns that are not independent from each other;
performing a greedy algorithm to determine one or more least-connected nodes, the one or more least-connected nodes having zero edges or the least number of edges to neighboring nodes in the graph among the nodes in the graph;
for individual of the one or more least-connected nodes:
collectively identifying the candidate semantic patterns associated with the individual least-connected nodes and the candidate semantic patterns associated with any neighboring nodes connected to the individual least-connected node in the graph as one of the final semantic patterns;
removing the individual least-connected node and any neighboring nodes connected to the individual least-connected node from the graph; and
assigning an independence ranking to the final semantic pattern; and
iteratively performing the greedy algorithm until the graph is empty, the independence rankings of the final semantic patterns identified in earlier iterations of the greedy algorithm being greater than independence ranking of final semantic patterns identified in later iterations of the greedy algorithm.

### Claim 22 (depends on 16)

The one or more computer-readable storage media of claim 16, the method further comprising:
receiving a set of SEM images taken of one or more wafers during manufacture of the one or more wafers; and
for individual of the SEM images, for individual of one or more areas of interest in the individual SEM images, generating one of the items in the continuous itemset.

### Claim 23 (depends on 22)

The one or more computer-readable storage media of claim 22, the method further comprising:
for the individual area of interest in the individual SEM images:
extracting one or more as-processed feature contours;
determining the presence or absence of the one or more manufacturing defects based at least in part on the individual SEM image; and
correlating the as-processed features contours to one or more correlated physical design feature values in a physical design;
wherein the physical design feature values for the item generated for the individual area of interest comprises the correlated physical design feature value and the event value for the item generated for the individual area of interest indicates the presence or absence of the one or more manufacturing defects at the individual area of interest.

### Claim 24

A semantic pattern extraction system comprising:
a discretization means for performing entropy-based discretization on a continuous itemset to generate a discretized itemset, individual of the items in the continuous itemset comprising a physical design feature value for one or more physical design features and an event value indicating the presence or absence of one or more manufacturing defects;
a mining means for identifying a set of candidate semantic patterns from the discretized itemset by performing frequent set mining on the discretized itemset;
a reduction means for reducing the set of candidate semantic patterns to one or more final semantic patterns;
a scoring means for determining a score for individual of the final semantic patterns; and
a display means to display the final semantic patterns in a ranked order, the ranked order based at least in part on the score for the individual final semantic patterns.

### Claim 25 (depends on 24)

The semantic pattern extraction system of claim 24, wherein the reducing comprises:
representing the candidate semantic patterns in a graph, wherein nodes in the graph are associated with individual of the candidate semantic patterns and edges in the graph connect nodes associated with candidate semantic patterns that are not independent from each other;
performing a greedy algorithm to determine one or more least-connected nodes, the one or more least-connected nodes having zero edges or the least number of edges to neighboring nodes in the graph among the nodes in the graph;
for individual of the one or more least-connected nodes:
collectively identifying the candidate semantic patterns associated with the individual least-connected nodes and the candidate semantic patterns associated with any neighboring nodes connected to the individual least-connected node in the graph as one of the final semantic patterns;
removing the individual least-connected node and any neighboring nodes connected to the individual least-connected node from the graph; and
assigning an independence ranking to the final semantic pattern; and
iteratively performing the greedy algorithm until the graph is empty, the independence rankings of the final semantic patterns identified in earlier iterations of the greedy algorithm being greater than independence ranking of final semantic patterns identified in later iterations of the greedy algorithm.