System and method for natural language processing and using ontological searches

• US 20060271353A1
• Filed: 05/30/2006
• Published: 11/30/2006
• Est. Priority Date: 05/27/2005
• Status: Active Grant

• Current Assignee: Hakia, Inc.
• Original Assignee: Hakia, Inc.
• Inventors: Raskin, Victor, Berkan, Riza C.
• Granted Patent: US 7,739,104 B2
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking

Abstract

A search engine is disclosed that utilizes natural language processing (NLP) techniques. The search engine utilizes meaning-based natural language processing using ontological semantics in analyzing the meaning of queries and the searched text. This system analyzes Web pages and queries. The NLP method produces equivalent meanings to a sequence of user initiated words, wherein relevance parsing of the original query produces a display of queries/questions as hot links to the next round of searching without additional typing by the user.

Claims

1. A method for data searching comprising:

collecting data utilizing a computer network;

parsing the data by language type;

expanding the data using ontological semantics;

categorizing the data and expanded data by anticipated queries;

storing the data and the expanded data based on anticipated queries;

receiving a query;

parsing and expanding the query and comparing the parsed user query with the anticipated queries; and

providing the matching data and the expanded data based in part on matching anticipated queries.

2. The method as recited in claim 1, wherein query is parsed into word groups and sentence groups.

3. The method as recited in claim 1, further comprising highlighting a portion of the matching data expanded data based on NLP analysis between query and data.

4. The method as recited in claim 1, further comprising selecting the best sentences to be displayed as a dialogue response.

5. The method as recited in claim 1, wherein the query is input using natural language.

6. The method as recited in claim 1, wherein the matching data and the expanded data includes a hyperlink to the original source of the data.

7. The method as recited in claim 1, further comprising:

scoring the data by adding a weight for each word and dividing by the sum of all of the words.

8. The method as recited in claim 7, wherein the scoring is calculated such that

$\mathrm{Sx}=\frac{\sum _{}}{}$

i = 1 , M ⁢

 

⁢

R i ⁢

W i ∑

i = 1 , M ⁢

 

⁢

W i ≤

1.0

9. The method as recited in claim 7, wherein the data includes URLs, page titles, and paragraphs.

10. The method as recited in claim 1, further comprising:

scoring the data based on the parsed data;

scoring the data based on a URL and title of the data; and

scoring the data based on the query.

11. The method as recited in claim 3, further comprising presenting the highlighted data in the form of response to the query.

12. The method as recited in claim 4, further comprising adding linguistic hedges to the best sentences to mimic dialog.

13. The method as recited in claim 10, wherein the URL scoring is calculated such that:

$S_{\mathrm{dom}}={\mathrm{Ss}<br><br>\left(\frac{J}{Z}\right)}^M+c$

14. The method as recited in claim 10, wherein the score is degraded by distance between data words and an order of the data words such that:

$\mathrm{St}=\underset{\underset{\underset{i\ne <br><br>j}{j=1,M}}{i=1,M}}{\sum <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{Sd}}_{i,j}<br><br>{\mathrm{So}}_{i,j}<br><br>{\mathrm{Sw}}_{i,j}$

15. The method as recited in claim 10, wherein relevance parsing of the original query produces a display of queries/questions as hot links to the next round of searching without additional queries.

16. A search method for a communication network, comprising:

receiving a query including search terms;

analyzing the query to determine a query type;

providing a hierarchical data structure containing a plurality of unique files, said files sorted by search terms, the search terms contained in the query data structure are arranged in different hierarchical levels, the hierarchical data structure containing links to at least one document;

comparing the analyzed query to the hierarchical data structure to perform a document search based on the search terms selected from the query data structure, and

outputting a found documents as search result.

17. A search system for a communication network, comprising:

a receiving unit adapted to receive a query;

a response unit adapted to check the query to determine if a predetermined response file exists

a question analyzed adapted to analyze the query and determine a query type;

a natural language processor adapted to process the query based in part on the query type; and

a response file containing links to documents responsive to the query.

18. The search system for a communication network according to claim 17, further comprising a first database including a plurality of predetermined response file.

19. The search system for a communication network according to claim 18, wherein the predetermined response files are selected from the group consisting of definitions, weather, sports, and conversions.

20. The search system for a communication network according to claim 17, further comprising a module adapted to highlight at least one sentence based on NLP of the query and the documents.

21. The search system for a communication network according to claim 20, further comprising a module adapted to select a best sentence from among the highlighted sentences.

22. The method as recited in claim 20, further comprising a module adapted to present the highlighted sentence is in the form of response to the query.

23. The method as recited in claim 21, further comprising a module adapted to add linguistic hedges to the best sentences to mimic dialog.