Schedule for 2017 Web Information Extraction and Retrieval Web IE & IR

(1)

Web IE & IR

Dr. Chia-Hui Chang

National Central University, Taiwan 2017/02/13

(2)

 Google Search Engine

 News Search

 Image Search

 Book Search

 ^Product

 Search on Map

 ^{Job Search}

 People Search

 Event Search

Search on the Web

(3)

(4)

(5)

(6)

 Topic Search or Deep Search or your choi

ce

 Learn how to crawl, extract data, and in

dex data for better search experience on

the Web

 Major topics:

 Information Extraction

 Template Pages & Free Texts

 Information Retrieval

Course Goal

(7)

Constructing a POI database from Web

7

Query Log

POI Search System

POI DB POI

Name+Address +Description

POI DB POI

Name+Address +Description

Query

Associated Information

Extraction

Address Extraction & Normalizatio

n

Query-based Crawler

POI Extraction

Address-bearing pages

POI Name

POI Name & Address Pairing

Addres s Keyword

s

POI Pair

POI Name Recognitio

n Address Pattern

Associated Information Summarization

(8)

Key Components

 ^Crawler:

 Yellow Page Web sites  Deep Web data extraction

 Query-based crawler for Address-Bearing Pages (A BP).

 POI Extraction from ABP

 Address extraction

 POI name extraction

 POI name and address pairing

 Description of a POI

 Associated information extraction

 Summarization of the associated information

8

(9)

• As a portal for mobile devices to replace APP s like Web search engines

• Categorization of Chatbots:

 Template-Driven Chatbots

 Retrieval-Based Chatbots

 Generative-based Chatbots

 Schedules of Chinese Subtask

 Task registration due: Apr 2017

 STC run submission deadline: May 2017

 Relevance assessments: Jun-Jul 2017

 Results released to participants: Sep 1 2017

 Participants draft papers due: Oct 1 2017

NTCIR – Short Text Conversa

tion

(10)

Retrieval-Based Chatbots (Closed)

(11)

Generative-based Methods (Op

en)

(12)

 5 Quiz: 25%

 3 Assignments: 45%

 Oral presentation: 10%

 Class involvement: 5%

 Ask at least 1 question every week

 Final project: 20%

 Paper submission: +3~5 bonus

 SIGIR, SAC WT, AI and the Web

Grading

(13)

 Stanford NLP-Dan Jurafsky & Chris

 18 - 1 - Introduction to Information Retrieval

 18 - 2 - Term-Document Incidence Matrices

 18 - 3 - The Inverted Index

 18 - 4 - Query Processing with the Inverted In dex

 18 - 5 - Phrase Queries and Positional Indexes