• Stars
    star
    2,638
  • Rank 17,345 (Top 0.4 %)
  • Language
    C++
  • License
    BSD 3-Clause "New...
  • Created over 7 years ago
  • Updated over 3 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

A Toolkit for Industrial Topic Modeling
logo

Build Status License

Familia 开源项目包含文档主题推断工具、语义匹配计算工具以及基于工业级语料训练的三种主题模型:Latent Dirichlet Allocation(LDA)、SentenceLDA 和Topical Word Embedding(TWE)。 支持用户以“拿来即用”的方式进行文本分类、文本聚类、个性化推荐等多种场景的调研和应用。考虑到主题模型训练成本较高以及开源主题模型资源有限的现状,我们会陆续开放基于工业级语料训练的多个垂直领域的主题模型,以及这些模型在工业界的典型应用方式,助力主题模型技术的科研和落地。(English)

News!!!

近期,我们在PaddleHub 1.8版本中上线了Familia中的LDA模型,根据数据集的不同,区分为lda_news、lda_novel和lda_webpage。

PaddleHub使用起来非常便捷,我们以lda_news的使用来进行例子介绍。

  1. 首先,在使用PaddleHub之前,需要先安装PaddlePaddle深度学习框架,更多安装说明请查阅飞桨快速安装

  2. 安装Paddlehub: pip install paddlehub

  3. lda_news模型安装:hub install lda_news

  4. 具体使用:

import paddlehub as hub

lda_news = hub.Module(name="lda_news")
jsd, hd = lda_news.cal_doc_distance(doc_text1="今天的天气如何,适合出去游玩吗", doc_text2="感觉今天的天气不错,可以出去玩一玩了")
# jsd = 0.003109, hd = 0.0573171

lda_sim = lda_news.cal_query_doc_similarity(query='百度搜索引擎', document='百度是全球最大的中文搜索引擎、致力于让网民更便捷地获取信息,找到所求。百度超过千亿的中文网页数据库,可以瞬间找到相关的搜索结果。')
# LDA similarity = 0.06826

results = lda_news.cal_doc_keywords_similarity('百度是全球最大的中文搜索引擎、致力于让网民更便捷地获取信息,找到所求。百度超过千亿的中文网页数据库,可以瞬间找到相关的搜索结果。')
# [{'word': '百度', 'similarity': 0.12943492762349573}, 
#  {'word': '信息', 'similarity': 0.06139783578769882}, 
#  {'word': '找到', 'similarity': 0.055296603463188265}, 
#  {'word': '搜索', 'similarity': 0.04270794098349327}, 
#  {'word': '全球', 'similarity': 0.03773627056367886}, 
#  {'word': '超过', 'similarity': 0.03478658388202199}, 
#  {'word': '相关', 'similarity': 0.026295857219683725}, 
#  {'word': '获取', 'similarity': 0.021313585287833996}, 
#  {'word': '中文', 'similarity': 0.020187103312009513}, 
#  {'word': '搜索引擎', 'similarity': 0.007092890537169911}]

更加具体的介绍和使用方法可以看这里:https://www.paddlepaddle.org.cn/hublist?filter=en_category&value=SemanticModel

应用介绍

Familia目前包含的主题模型的对应论文介绍可以参考相关论文

主题模型在工业界的应用范式可以抽象为两大类: 语义表示和语义匹配。

  • 语义表示 (Semantic Representation)    对文档进行主题降维,获得文档的语义表示,这些语义表示可以应用于文本分类、文本内容分析、CTR预估等下游应用。

  • 语义匹配 (Semantic Matching)

    计算文本间的语义匹配度,我们提供两种文本类型的相似度计算方式:

    • 短文本-长文本相似度计算,使用场景包括文档关键词抽取、计算搜索引擎查询和网页的相似度等等。
    • 长文本-长文本相似度计算,使用场景包括计算两篇文档的相似度、计算用户画像和新闻的相似度等等。

更详细的内容及工业界应用案例可以参考Familia Wiki , 如果想要对上述应用范式进行基于Web的可视化展示,可以参考Familia-Visualization

代码编译

第三方依赖包括gflags-2.0glogs-0.3.4protobuf-2.5.0, 同时要求编译器支持C++11, g++ >= 4.8, 兼容Linux和Mac操作系统。 默认情况下执行以下脚本会自动获取依赖并安装。

$ sh build.sh # 包含获取并安装第三方依赖的过程

模型下载

$ cd model
$ sh download_model.sh

我们会陆续开放不同领域的多种主题模型,来满足更多不同的场景需求。

Demo

Familia自带的Demo包含以下功能:

  • 语义表示计算 利用主题模型对输入文档进行主题推断,以得到文档的主题降维表示。

  • 语义匹配计算   计算文本之间的相似度,包括短文本-长文本、长文本-长文本间的相似度计算。

  • 模型内容展现    对模型的主题词,近邻词进行展现,方便用户对模型的主题有直观的理解。

具体的Demo使用说明可以参考使用文档

注意事项

  • 若出现找不到libglog.so, libgflags.so等动态库错误,请添加third_party至环境变量的LD_LIBRARY_PATH中。

    export LD_LIBRARY_PATH=./third_party/lib:$LD_LIBRARY_PATH

  • 代码中内置简易的FMM分词工具,只针对主题模型中出现的词表进行正向匹配。若对分词和语义准确度有更高要求,建议使用商用分词工具,并使用自定义词表的功能导入主题模型中的词表。

问题咨询

欢迎提交任何问题和Bug Report至Github Issues。 或者发送咨询邮件至{ familia } at baidu.com

Docker

docker run -d \
    --name familia \
    -e MODEL_NAME=news \
    -p 5000:5000 \
    orctom/familia

MODEL_NAME can be one of news/novel/webpage/webo

API

http://localhost:5000/swagger/

Citation

The following article describes the Familia project and industrial cases powered by topic modeling. It bundles and translates the Chinese documentation of the website. We recommend citing this article as default.

Di Jiang, Yuanfeng Song, Rongzhong Lian, Siqi Bao, Jinhua Peng, Huang He, Hua Wu. 2018. Familia: A Configurable Topic Modeling Framework for Industrial Text Engineering. arXiv preprint arXiv:1808.03733.

@article{jiang2018familia,
  author = {Di Jiang and Yuanfeng Song and Rongzhong Lian and Siqi Bao and Jinhua Peng and Huang He and Hua Wu},
  title = {{Familia: A Configurable Topic Modeling Framework for Industrial Text Engineering}},
  journal = {arXiv preprint arXiv:1808.03733},
  year = {2018}
}

Further Reading: Federated Topic Modeling

Copyright and License

Familia is provided under the BSD-3-Clause License.

More Repositories

1

amis

前端低代码框架,通过 JSON 配置就能生成各种页面。
TypeScript
17,235
star
2

uid-generator

UniqueID generator
Java
5,429
star
3

san

A fast, portable, flexible JavaScript component framework
JavaScript
4,708
star
4

lac

百度NLP:分词,词性标注,命名实体识别,词重要性
C++
3,864
star
5

braft

An industrial-grade C++ implementation of RAFT consensus algorithm based on brpc, widely used inside Baidu to build highly-available distributed systems.
C++
3,499
star
6

dperf

dperf is a DPDK based 100Gbps network performance and load testing software.
C
3,273
star
7

bfs

The Baidu File System.
C++
2,853
star
8

openrasp

🔥Open source RASP solution
C++
2,774
star
9

AnyQ

FAQ-based Question Answering System
C++
2,584
star
10

sofa-pbrpc

A light-weight RPC implement of google protobuf RPC framework.
C++
2,130
star
11

Senta

Baidu's open-source Sentiment Analysis System.
Python
1,889
star
12

tera

An Internet-Scale Database.
C++
1,887
star
13

bfe-book

In-depth Understanding of BFE《深入理解BFE》(Book for BFE, a CNCF open source project. both in English and in Chinese)
1,212
star
14

BaikalDB

BaikalDB, A Distributed HTAP Database.
C++
1,169
star
15

bigflow

Baidu Bigflow is an interface that allows for writing distributed computing programs and provides lots of simple, flexible, powerful APIs. Using Bigflow, you can easily handle data of any scale. Bigflow processes 4P+ data inside Baidu and runs about 10k jobs every day.
C++
1,142
star
16

DuReader

Baseline Systems of DuReader Dataset
Python
1,133
star
17

DDParser

百度开源的依存句法分析系统
Python
973
star
18

starlight

Java implementation for Baidu RPC, multi-protocol & high performance RPC.
Java
961
star
19

CUP

CUP, common useful python-lib. (Currently, Most popular python lib in baidu). Python 开发底层库, 涵盖util、service(threadpool/generator/executor/cache等等)、logging、monitoring、增强型配置 等等库支持
Python
938
star
20

ICE-BA

C++
700
star
21

NoahV

An efficient front-end application framework based on vue.js
JavaScript
639
star
22

EasyFaaS

EasyFaaS是一个依赖轻、适配性强、资源占用少、无状态且高性能的函数计算服务引擎
Go
620
star
23

Curve

An Integrated Experimental Platform for time series data anomaly detection.
JavaScript
530
star
24

Jprotobuf-rpc-socket

Protobuf RPC是一种基于TCP协议的二进制RPC通信协议的Java实现
Java
516
star
25

bifromq

A MQTT broker implementation adopting serverless architecture
Java
514
star
26

fast_rgf

Multi-core implementation of Regularized Greedy Forest
C++
466
star
27

babylon

High-Performance C++ Fundamental Library
C++
457
star
28

Dialogue

Python
444
star
29

Elasticsearch

Baidu Elasticsearch
Java
432
star
30

brcc

BRCC(better remote config center)是一个分布式配置中心,用于统一管理应用服务的配置信息,避免各类资源散落在各个项目中,简化资源配置的维护成本。作为一种轻量级的解决方案,部署简单,同时支持多环境、多版本、多角色的资源管理,可以在不改变应用源码的情况下无缝切换和实时生效配置信息。
Java
390
star
31

Cafe

A powerful test framework for Android
Java
370
star
32

mix-img

A fast mix image javascript tool libary
JavaScript
332
star
33

puck

Puck is a high-performance ANN search engine
Jupyter Notebook
331
star
34

unit-dmkit

C++
327
star
35

galaxy

Galaxy is a cluster management system.
C++
326
star
36

information-extraction

Python
325
star
37

knowledge-driven-dialogue

baseline system of knowledge driven dialogue competition
Python
270
star
38

CarbonGraph

A Swift dependency injection / lookup framework for iOS
Swift
254
star
39

unit-uskit

unit-uskit
C++
251
star
40

BIPlatform

JavaScript
219
star
41

dlock

An effective and reliable Distributed Lock
Java
216
star
42

ins

iNexus, coordinate large scale services
C++
214
star
43

boteye

C++
212
star
44

titan-dex

Java
201
star
45

m-git

MGit 是一款基于 Git 的多仓库管理工具,可以安全的、高效的管理多个 Git 仓库; 适合于在多个仓库中进行关联开发的项目,实现批量的版本管理功能,提高 Git 操作的效率,避免逐个执行 Git 命令带来的误操作风险。
Ruby
166
star
46

Rubik

An Android platform component management tool chain, based on Kotlin language.
Kotlin
154
star
47

common

Common library
C++
132
star
48

go-lib

Go
126
star
49

titan-hotfix

Java
125
star
50

wx2

小程序互转工具
JavaScript
124
star
51

iot-sdk-c

device sdk for baidu IoT Core service, in c. Including MQTT client
C
118
star
52

Youtube-8M

PaddlePaddle models for Youtube-8M Video Understanding Challenge
Python
114
star
53

ar-sdk

DuMix AR SDK for Developer
GLSL
107
star
54

broc

Python
101
star
55

ITEST

Web service interface test framework
97
star
56

ote-stack

OTE-Stack is an edge computing platform for 5G and AI
Go
96
star
57

GPT

Java
87
star
58

redis

Baidu Ksarch Redis - a production solution of redis cluster
87
star
59

san-devtools

Browser developer tools extension for debugging San.
TypeScript
82
star
60

terminator

Service Virtualization
Java
76
star
61

QCompute

QCompute is a Python-based quantum software development kit (SDK). It provides a full-stack programming experience for advanced users via hybrid quantum programming language features and a high-performance simulator.
Python
76
star
62

spring-cloud-baidu

70
star
63

shuttle

A fast computing framework based on Galaxy
C++
64
star
64

iot-edge-sdk-for-iot-parser

C
64
star
65

baidu-iot-samples

C
61
star
66

san-store

Application States Management for San
JavaScript
59
star
67

ARK

Development framework of intelligent operation
Python
57
star
68

san-update

Object immutable update utility for san solution
JavaScript
56
star
69

logcover

轻量级异常日志测试覆盖率度量工具
Python
56
star
70

palo

A fast MPP database for all modern analytics on big data. Powered by Apache Doris(Incubating)
50
star
71

speech-samples

百度语音示例
Java
48
star
72

ntripcaster

C
43
star
73

san-router

Official Router for San
JavaScript
38
star
74

Quanlse

Jupyter Notebook
38
star
75

san-ssr

San SSR framework and utils
TypeScript
37
star
76

dm-kit-php

PHP
36
star
77

boteye_sensor

C
35
star
78

ipipe-agent

Java
33
star
79

OASP

OASP (Online App Status Protocol)
Java
32
star
80

san-composition

JavaScript
30
star
81

duedge-recipes

DuEdge百度边缘网络计算样例代码
JavaScript
27
star
82

paddle-on-k8s-operator

Kubernetes operator for managing the lifecycle of PaddlePaddle job.
Go
24
star
83

baiducloud-sdk-go

Go SDK for Baidu Cloud
Go
24
star
84

san-website

JavaScript
21
star
85

baiduads-sdk

Baidu Ads API SDK
Python
19
star
86

du1906_esp

DUHOME AIOT platform based on du1906 and esp32
C
18
star
87

highflip

HIGHFLIP: An easy way to bridge different federal learning platforms
18
star
88

smartapp-openapi-java

百度智能小程序服务端 OpenAPI SDK for java,是基于小程序服务端 OpenAPI 封装的一套让开发者方便使用的 SDK, 它可以帮开发者减少理解和使用 OpenAPI 的成本, 减少开发者直接调用服务端接口不当而引起的错误, 避免在开发中走弯路。
Java
16
star
89

san-factory

JavaScript
15
star
90

ttm

C
14
star
91

cluster-api-provider-baiducloud

Kubernetes cluster-api for Baidu Cloud
Go
13
star
92

minions

Baidu 100G Chasiss Switch hardware spec
11
star
93

signet

签章系统
JavaScript
10
star
94

sgxray

SGXRay: a bounded verifier for Intel SGX enclaves
C
10
star
95

grafana-tsdb-datasource

JavaScript
9
star
96

iotcore-sdk-java

Java SDK for baidu IoT Core service
Java
9
star
97

bce-fpga-dev-kit

VHDL
8
star
98

iot

for all code about Internet of Things
8
star
99

smartapp-openapi-go

百度智能小程序服务端 OpenAPI SDK for go,是基于小程序服务端 OpenAPI 封装的一套让开发者方便使用的 SDK, 它可以帮开发者减少理解和使用 OpenAPI 的成本, 减少开发者直接调用服务端接口不当而引起的错误, 避免在开发中走弯路。
Go
8
star
100

duedge-cli

DuEdge Command Line
Python
6
star