# Introduction to IR-IE

###### tags: `Information Retrieval and Extraction`

## Text Information Retrieval

### Architecture of IR Systems

### Information Retrieval Models
* Boolean model [[1]](/rCAXfaG0RwKSY2mJiP8ZeQ)
* Vector model [[2]](/B6RlFoXdTjK23MVw5Yy6cw)
* Probabilistic model [[3]](/49SxQomqSYmqMMy24RzZFg)
* Language Model [[4]](/c3mb4r6aTQSMljysMuYM9g)
* Topic Model [[5]](/Y-qNVayLTAeovtUSjqlfsw)
* Deep Learning Model [[6]](/H6yyQNSJSkmOh92YhfDGMQ)


## Multilingual Information Retrieval

### Multi- & Cross- Lingual Information Access

### Enhancing Traditional Information Retrieval Systems


(1): text translation
(2): vector translation
(3): query translation
(4): term vector translation
(1) and (2), (3) and (4): interlingual form

## Multimedia Information Retrieval

### Semantic Gap

以前從圖像中去擷取隱含的Semantic content是有困難的，但現在由於CV技術進步，處理能力提高

### Image Retrieval Black Box

## Web Information Retrieval

#### The big challenge Meet the user needs given the heterogeneity of web pages
從龐大的資料中萃取所需資訊是很困難的，像是網站的一直性（heterogeneity of web pages），且會根據不同背景的人，輸入的搜尋關鍵字(poorly made queries)也會不同（成年人、小孩）

## Context-Aware Information Retrieval

### Definitions of Context for IR

* IR: location and delivery of documents which satisfy a user information need.
* IR takes place in “context”, but this context is generally ignored in IR models and system design.
* The definition of context in IR is widely interpreted.

## Information Extraction

Automatic extraction of structured information from unstructured sources
* entities
* relationships between entities
* attributes describing entities

### What is “Information Extraction”
Information Extraction = segmentation + classification + clustering + association


## Information Extraction vs. Information Retrieval

哪一個要先做其實不一定，像是ＱＡ問題會先做IR > IE

## Search Scenario


### Elements in the Search Scenario
* Users
* Users of Different Backgrounds, Expression , Contexts Naïve Users or Expert Users
* Queries
* Gaps between Information need and queries
* Same information need expressed in different queries
* Manual vs. Automatic (term suggestion, concept recommendation, relevance feedback)

### Elements in the Retrieval Scenario
* IR Systems
* Evaluation Metrics
* Results
* 檢索結果會根據不同人的認定會有不同的好或壞