Systems and methods for website improvement

• US 10,558,718 B2
• Filed: 11/03/2015
• Issued: 02/11/2020
• Est. Priority Date: 11/03/2015
• Status: Active Grant

• Current Assignee: Dell Products LP (Dell Technologies Inc.)
• Original Assignee: Dell Products LP (Dell Technologies Inc.)
• Inventors: Cerveny, Lukas, De Jaegher, Arnaud
• Primary Examiner(s): Bullock, Joshua
• Assistant Examiner(s): Bui, Tiffany Thuy
• Application Number:
  US14/931,095
• Publication Number:
  US 20170124192A1
• Time in Patent Office: 1,561 Days
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation

Abstract

A method for using evaluation of intent to improve website usability includes gathering page text and input text from pages viewed by a user, generating a word set from the page text and input text, and performing latent dirichlet allocation modeling on the word set to evaluate user intent. The intent can be used to provide recommendations, improve search results, or identify weaknesses in a website design.

Claims

1. A method for using evaluation of intent to improve an online support website, comprising:

storing in computer memory page text and input text from web pages viewed by a user;

using a processor to generate a word set from the page text and input text;

generating a total number of occurrences of a particular word in the page text and input text by subtracting negative scored words for the particular word from other occurrences of the particular word within the page text and input text;

storing the total number of occurrences of the particular word in the word set;

transforming the word set to removing stop words and to perform word transformations;

normalizing nouns to one form and verbs to a single tense, person, and voice;

performing latent dirichlet allocation modeling on the word set to evaluate user intent, wherein the evaluation of user intent includes determining one or more user intents based on the latent dirichlet allocation modeling;

redetermining intent after additional page visits and refining the recommendations based on the redetermined intent, wherein the redetermined intent includes identifying topics associated with the additional page visits and the refined recommendations are based on the topics;

ranking the one or more user intents according to a probability of the one or more user intents;

selecting one of the user intents based on the ranking of the one or more user intents; and

providing support pages likely to help with a problem the user is experiencing to the user based on the user intent.

2. The method of claim 1, wherein the page text includes page titles, text on a page, keywords, page metadata, or any combination thereof.

3. The method of claim 1, wherein the recommendations are ranked by a correlation score, an intent score, or any combination thereof.

4. A method for using evaluation of intent to improve website usability, comprising:

using a processor to gather page text and input text from web pages viewed by a user;

generating a word set from the page text and input text;

performing latent dirichlet allocation modeling on the word set to evaluate user intent;

correlating pages visited with the user intent;

identifying off-intent navigations and return-to-intent navigations, wherein off-intent navigations are one or more pages visited that are not correlated with the user intent;

calculating a duration of the off-intent navigations based on the number of pages viewed between an off-intent navigation and a corresponding return-to-intent navigation;

scoring a site design based on frequency of off-intent navigations and duration of off-intent navigations;

in response to the scoring of the site design, identifying links that led to a particular number of off-intent navigations;

modifying the identified links to improve the scoring of the site design;

tracking the site design score as changes are made to the site design or comparing the site design score across multiple site designs; and

writing results of the scoring to computer memory.

5. The method of claim 4, further comprising transforming the word set to removing stop words, to normalize nouns to one form and verbs to a single tense, person, and voice, and to perform word transformations.

6. The method of claim 4, wherein the page text includes page titles, text on a page, keywords, page metadata, or any combination thereof.

7. The method of claim 4, wherein the user intent include more than one intent scored by a likelihood or probability of the intent.