Automated routing of voice calls using time-based predictive clickstream data

Patent metadata

Field	Value
Publication number	US-9800727-B1
Application number	US-201615294472-A
Country	US
Kind code	B1
Filing date	Oct 14, 2016
Priority date	Oct 14, 2016
Publication date	Oct 24, 2017
Grant date	Oct 24, 2017

Abstract

Methods and apparatuses are described for automated routing of voice calls using time-based predictive clickstream data. A server captures clickstream data comprising uniform resource locators (URLs) and one or more timestamps of a web session. The server converts the clickstream data into tokens and generates a frequency matrix based upon the tokens. The server generates a feature vector based upon the frequency matrix. The server receives an incoming voice call from a remote device and identifies that the remote device is associated with a user of the client computing device. The server determines intent for the incoming voice call based upon the feature vector, and routes the incoming voice call to a destination device based upon the determined intent.

First claim

What is claimed is:

1. A method for automated routing of voice calls using time-based predictive clickstream data, the method comprising:
capturing, by a server computing device at a first point in time, clickstream data corresponding to one or more web browsing sessions between a client computing device and a web server, the clickstream data comprising uniform resource locators (URLs) and one or more timestamps of the corresponding session;
converting, by the server computing device, the clickstream data into tokens, comprising filtering the URLs to retain intent-relevant URLs;
parsing each intent-relevant URLs into one or more tokens, each token comprising a discrete text segment of the corresponding URL;
assigning a time value to each token that is associated with at least one of the timestamps from the corresponding web browsing session;

generating, by the server computing device, a frequency matrix based upon the tokens, the frequency matrix comprising for each token and web browsing session (i) a frequency of the token appearing in the intent-relevant URLs in the session (“TF”) and (ii) a log transform of an inverse of a ratio of number of distinct intent-relevant URLs that include the token over the number of intent-relevant URLs in the session (“IDF”);
generating, by the server computing device, a feature vector based upon the frequency matrix, the feature vector comprising for each token a value indicating a product of TF and IDF;
receiving, by the server computing device at a second point in time, an incoming voice call from a remote device;
identifying, by the server computing device, that the remote device is associated with a user of the client computing device;
determining, by the server computing device, intent for the incoming voice call based upon the feature vector; and
routing, by the server computing device, the incoming voice call to a destination device based upon the determined intent.

2. The method of claim 1 , wherein the step of filtering the URLs comprises removing URLs from the clickstream data that are previously determined to be irrelevant to intent.

3. The method of claim 1 , wherein the step of converting the clickstream data into tokens further comprises
determining one or more tags embedded in a webpage that corresponds to one of the intent-relevant URLs; and
capturing the tags as part of the clickstream data.

4. The method of claim 1 , wherein the URLs include one or more search keywords, the method further comprising correcting errors in the search keywords by comparing the search keywords against a predefined list of keywords.

5. The method of claim 1 , wherein the step of converting the clickstream data into tokens further comprises
identifying one or more alphanumeric codes in the intent-relevant URLs that correspond to a financial security; and
inserting other information relating to the financial security into the intent-relevant URL.

6. The method of claim 5 , wherein the alphanumeric codes are CUSIP numbers.

7. The method of claim 1 , wherein the step of parsing each intent-relevant URLs into one or more tokens comprises
removing generic and user-specific portions of each intent-relevant URL based upon a predefined syntax;
identifying one or more non-alphanumeric characters in the URL;
splitting each intent-relevant URL into sections based upon a position of the non-alphanumeric characters in the URL;
removing the non-alphanumeric characters from the URL; and
identifying one or more tokens within each section.

8. The method of claim 1 , wherein the time value assigned to each token is a past window of time during which the associated web browsing session occurred.

9. The method of claim 1 , wherein the step of determining intent for the incoming voice call comprises weighting each token in the feature vector according to the assigned time value of the token.

10. The method of claim 9 , wherein tokens with an assigned time value that is closer to the second point in time are given more weight than tokens with an assigned time value that is farther from the second point in time.

11. The method of claim 1 , wherein the step of determining intent for the incoming voice call comprises comparing tokens in the feature vector with a predefined list of intents and selecting an intent from the predefined list of intents that matches the tokens in the feature vector.

12. The method of claim 1 , wherein the step of identifying that the remote device is associated with a user of the client computing device comprises
receiving a device identifier from the remote device;
determining an identity of a user associated with the remote device based upon the device identifier; and
determining that the identity of the user is also associated with the client computing device.

13. The method of claim 1 , wherein the step of identifying that the remote device is associated with a user of the client computing device comprises
receiving a user identifier from the remote device;
determining an identity of a user associated with the remote device based upon the user identifier; and
determining that the identity of the user is also associated with the client computing device.

14. A system for automated routing of voice calls using time-based predictive clickstream data, the system comprising a server computing device configured to:
capture, at a first point in time, clickstream data corresponding to one or more web browsing sessions between a client computing device and a web server, the clickstream data comprising uniform resource locators (URLs) and one or more timestamps of the corresponding session;
convert the clickstream data into tokens, comprising filtering the URLs to retain intent-relevant URLs;
parsing each intent-relevant URLs into one or more tokens, each token comprising a discrete text segment of the corresponding URL;
assigning a time value to each token that is associated with at least one of the timestamps from the corresponding web browsing session;

generate a frequency matrix based upon the tokens, the frequency matrix comprising for each token and web browsing session (i) a frequency of the token appearing in the intent-relevant URLs in the session (“TF”) and (ii) a log transform of an inverse of a ratio of number of distinct intent-relevant URLs that include the token over the number of intent-relevant URLs in the session (“IDF”);
generate a feature vector based upon the frequency matrix, the feature vector comprising for each token a value indicating a product of TF and IDF;
receive, at a second point in time, an incoming voice call from a remote device;
identify that the remote device is associated with a user of the client computing device;
determine intent for the incoming voice call based upon the feature vector; and
route the incoming voice call to a destination device based upon the determined intent.

15. The system of claim 14 , wherein filtering the URLs comprises removing URLs from the clickstream data that are previously determined to be irrelevant to intent.

16. The system of claim 14 , wherein converting the clickstream data into tokens further comprises
determining one or more tags embedded in a webpage that corresponds to one of the intent-relevant URLs; and
capturing the tags as part of the clickstream data.

17. The system of claim 14 , wherein the URLs include one or more search keywords, the server computing device configured to correct errors in the search keywords by comparing the search keywords against a predefined list of keywords.

18. The system

Assignees

• Fmr Llc

Inventors

• Chakrabarty Bibhash
• Arias-Vargas Alex

Classifications

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06Q30/0201

  Market modelling; Market analysis; Collecting market data · CPC title

• H04M3/5232 Primary

  Call distribution algorithms · CPC title

• G06F17/30477

  Physics · mapped topic

• G06N7/005

  Physics · mapped topic