# US 2021/0344636 A1 · Intelligent transaction scoring

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

- Number: US 2021/0344636 A1
- Document ID: US20210344636A1
- Status: Published application
- Kind: A1
- Application number: 16/928,397
- Filed: 2020-07-14
- Published: 2021-11-04
- Earliest claimed: 2020-04-29
- Assignees: Clarabridge, Inc.
- Applicants: Clarabridge, Inc.
- Inventors: Fabrice Martin; Ellen Loeshelle; Keegan Brenneman; Ram Ramachandran; Maksym Shcherbina; Kenneth Voorhees; Ramy Zulficar
- Web page: https://patentgrep.com/patent/US20210344636A1

## CPC

- G06F 40/253: Grammatical analysis; Style critique
- H04L 51/52: for supporting social networking services
- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06F 40/30: Semantic analysis
- G06F 40/35: Discourse or dialogue representation
- G06N 5/041: {Abduction}
- G06N 20/00: Machine learning
- H04L 51/18: Commands or executable codes
- H04L 51/234: for tracking messages
- H04L 51/42: Mailbox-related aspects, e.g. synchronisation of mailboxes
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods provide a flexible environment for intelligently scoring transactions from data sources. An example method includes providing a user interface configured to generate a rubric by receiving selection of one or more one or more conditions, each condition identifying a tag generated by a classifier in the library of classifiers and for each identified tag, receiving selection of a value for the tag that satisfies the condition and receiving selection of an outcome attribute for the condition. The outcome attribute may be a weight for the tag or an alert condition. The method includes storing the rubric in a data store and applying the stored rubric to scoring units of a transaction. The method also includes aggregating scores for transactions occurring during a trend period and displaying the trend score. In some implementations, at least one classifier in the library is a rule-based classifier defined by a user.

## Claims

### Claim 1

A system for intelligently scoring transactions from a data source, the system comprising:
at least one processor;
memory storing a library of classifiers;
memory storing instructions that, when executed by the at least one processor, causes the system to perform operations including:
providing a user interface configured to generate a rubric by:
receiving selection of one or more scoring conditions, each scoring condition identifying a tag generated by a classifier in the library of classifiers and
for each identified tag, receiving selection of a value for the tag that satisfies the scoring condition and receiving selection of an outcome attribute for the scoring condition, and
storing the rubric in a data store; and
memory storing a scoring engine configure to apply the stored rubric to scoring units of a transaction to generate a score for the transaction,
wherein the operations further include providing a dashboard interface that displays a trend score, the trend score being an aggregate of scores for transactions occurring during a trend period.

### Claim 2 (depends on 1)

The system of claim 1, wherein the outcome attribute includes a weight, the weight being used to generate the score for a transaction when the transaction satisfies the scoring condition.

### Claim 3 (depends on 1)

The system of claim 1, wherein the outcome attribute includes an alert condition.

### Claim 4 (depends on 3)

The system of claim 3, wherein the scoring engine is further configured to place a transaction that meets the alert condition on an alert queue and the system further comprises memory storing instructions that, when executed by the at least one processor, cause the system to provide an alert interface configured to, for a set of transactions in the alert queue:
display an identifier for each transaction in the set of transactions;
display a snippet from each transaction in the set of transactions; and
display, for each transaction in the set of transactions, tags generated for the transaction by application of the scoring engine.

### Claim 5 (depends on 1)

The system of claim 1, wherein at least one classifier is a rule-based classifier defined by a user.

### Claim 6 (depends on 5)

The system of claim 5, wherein the user interface is further configured to receive the rule-based classifier from a user, wherein receiving the rule-based classifier includes receiving a particular condition, the rule-based classifier testing for presence or absence of the particular condition in the transaction, and receiving a tag for the rule-based classifier.

### Claim 7 (depends on 1)

The system of claim 1, wherein the value for the tag includes an indication of presence of the tag or an indication of absence of the tag.

### Claim 8 (depends on 1)

The system of claim 1, wherein the library of classifiers includes an emotional intensity classifier, wherein a tag generated by the emotional intensity classifier is based on a confidence in a prediction that a scoring unit includes an intense emotion.

### Claim 9 (depends on 1)

The system of claim 1, wherein the user interface is further configured to receive rubric selection criteria that is used to select transactions from data sources.

### Claim 10 (depends on 1)

The system of claim 1, wherein the dashboard interface is further configured to:
display a trend graph, the trend graph breaking the trend score into smaller aggregations over the trend period.

### Claim 11 (depends on 1)

The system of claim 1, wherein the trend score is a first trend score and the dashboard interface is further configured to:
display a trend score for each of at least two attribute values of a transaction attribute, wherein the first trend score is for one of the at least two attribute values.

### Claim 12 (depends on 11)

The system of claim 11, wherein the dashboard interface is further configured to:
display a trend graph for each trend score, the trend graph breaking the trend score into smaller aggregations over the trend period.

### Claim 13 (depends on 11)

The system of claim 11, wherein the dashboard interface is further configured to receive a selection of the transaction attribute from a user.

### Claim 14 (depends on 1)

The system of claim 1, wherein the user interface is further configured to:
receive a group label from a user; and
enable linking of two or more tags under the group label, wherein the two or more tags each receives a weight as the outcome attribute and wherein a weight for the group label is an aggregate of the weights for the two or more tags.

### Claim 15 (depends on 1)

The system of claim 1, wherein the user interface is further configured to:
receive an indication that one of the selected tags is a branching criterion; and
enable linking of two or more tags under the branching criterion.

### Claim 16

A method comprising:
generating a repository of scored transactions by, for each transaction of a plurality of transactions:
determining that a customized rubric applies to the transaction based on selection criteria for the customized rubric, the customized rubric including scoring criteria, the scoring criteria identifying a plurality of tags, each tag being a tag generated by a classifier from a library of classifiers and at least some of the tags having a weight associated with a tag value,
for each scoring unit of the transaction, applying the classifiers associated with the plurality of tags to the scoring unit, wherein applying the classifiers labels the scoring unit with tags,
aggregating the tags of the scoring unit to generate transaction level tags,
calculating a transaction score based on the transaction level tags and the scoring criteria, and
storing the transaction score and the transaction level tags in the repository; and
displaying a dashboard that includes a trend score, the trend score being an aggregate of transaction scores for transactions occurring during a trend period.

### Claim 17 (depends on 16)

The method of claim 16, wherein the transactions include social media posts and the transaction scores are calculated in real-time.

### Claim 18

A method comprising:
obtaining labeled scoring units, each labeled scoring unit being labeled as including an emotion or not including an emotion and for scoring units labeled as including an emotion, being labeled as reflecting emotional intensity or not reflecting emotional intensity;
training, using the labeled scoring units, a machine-learned classifier to predict an emotional intensity label;
using the trained machine-learned classifier to predict an emotional intensity label for a new transaction scoring unit, the prediction including a confidence level; and
adding an emotional intensity tag to the scoring unit based on a value of the confidence level.

### Claim 19 (depends on 18)

The method of claim 18, wherein adding an emotional intensity tag to the scoring unit based on a value of the confidence level includes:
adding a low intensity tag to the scoring unit when the confidence level falls in a low range;
adding a high intensity tag to the scoring unit when the confidence level falls in a high range; and
adding a medium intensity tag to the scoring unit when the confidence level falls outside the low range and the high range.

### Claim 20 (depends on 18)

The method of claim 18, wherein the machine-learned classifier is one classifier of a plurality of classifiers in a library and the method further comprises:
providing a user interface configured to generate a rubric by:
receiving selection of one or more scoring conditions, each scoring condition identifying a tag generated by a classifier in the library of classifiers, at least one of the one or more scoring conditions identifying the emotional intensity tag,
for each identified tag, receiving selection of a value for the tag that satisfies the scoring condition, and
for at least some of the identified tags, receiving a weight for the tag;
applying the rubric to scoring units of a plurality of transactions, wherein applying the rubric to a transaction of the plurality of transactions generates a score for the transaction; and
displaying a trend score, the trend score being an aggregate of scores for transactions occurring during a trend period.
