[ADW24-D1-S06] AI-Powered Tech Doc – Content Quality with Acrolinx (Michael Klemme, Ralf Kühnlein)

Adobe Technical Communication
13 Jun 202438:01

Summary

TLDR本视频介绍了Acrin公司如何利用人工智能助手提高企业内容质量。Acrin通过集成到作者喜爱的创作应用中,根据企业写作标准自动检查和评分内容,提供侧边栏问题提示和AI生成的内容改进建议。此外,Acrin AI助手还能帮助作者生成符合企业标准的新内容,如最佳实践。演示了在Adobe Experience Manager Guides中的应用,展示了如何通过Acrin提升内容清晰度、一致性和可读性,并使用术语管理功能确保技术术语的统一性。Acrin还展示了其内容分析工具Content Cube,它可以扫描已发布网页,确保内容与企业风格指南一致。最后,介绍了Acrin的生成性AI功能,它能够重写复杂句子,为作者提供更易读的内容选项。

Takeaways

  • 😀 Acrin 提供了一个 AI 助手,帮助作家在他们喜欢的创作应用中检查内容质量,并与企业建立的写作标准进行比较,提供质量评分。
  • 🔍 Acrin 会将作家的内容与企业标准进行比对,并在侧边栏中展示问题,供作家点击修正,以提高内容的一致性。
  • 📝 Acrin 提供 AI 生成的内容改进建议,这些建议与企业写作标准一致,帮助快速提升内容质量。
  • 🎼 当作家需要帮助生成新内容时,可以启动 AI 助手,它能够根据企业批准的内容生成标准提供帮助。
  • 🔄 Acrin 允许作家在内容生成后,通过 AI 助手窗口进行比较或将其保存以便日后快速查找。
  • 🛠️ Acrin 集成到内容创作过程中,通过侧边栏直接集成到工作场所,帮助作者在写作时直接纠正形式问题。
  • 📈 Acrin 还包括分析数据库,帮助企业做出关于内容整体质量的决策,并提供改进机会。
  • 👥 Acrin 适用于不同团队,包括技术文档编写者、支持人员等,帮助他们控制发送给最终用户的内容。
  • 🚦 Acrin 可以设置质量门,确保内容在进入下一步骤前达到一定的质量水平。
  • 🌐 Acrin 还包括 AC 内容立方体,可以扫描整个公开网页,帮助企业监控和改进已发布内容的一致性。
  • 📝 Acrin 正在开发新的功能,如使用生成性 AI 来重写复杂句子,提供更易于理解的句子建议。

Q & A

  • ACRIN是什么,它如何帮助作家提高内容质量?

    -ACRIN是一个AI助手,它通过与企业写作标准比较,为作家提供内容质量评分和改进建议,帮助他们在喜欢的写作应用程序中检查内容质量。

  • ACRIN如何展示内容问题?

    -ACRIN在侧边栏展示内容问题,作家可以点击这些问题进行修正,以符合内容对齐、清晰度等写作标准。

  • ACRIN提供哪些写作标准的反馈?

    -ACRIN提供关于清晰度、术语、包容性、正确性、可扫描性、一致性和风格等多个写作标准的反馈。

  • 超过多少百分比的内容问题提供了可点击的指导?

    -超过87%的内容问题提供了可点击的指导,这大幅提高了作家的效率。

  • ACRIN AI助手如何帮助作家生成新内容?

    -当作家需要帮助生成新内容时,他们可以从喜欢的应用程序启动AI助手,AI助手会根据请求提供研究和答案。

  • ACRIN AI助手生成的内容如何与企业写作标准对齐?

    -ACRIN AI助手生成的内容会根据企业的写作标准进行定制,确保内容符合企业标准并获得评分。

  • ACRIN如何帮助作家更有效地使用术语?

    -ACRIN通过术语管理功能,帮助作家使用一致的术语,确保整个文档中技术术语的一致性。

  • ACRIN的术语管理功能如何帮助处理公司产品重命名的情况?

    -ACRIN的术语管理功能允许快速更新和替换术语,以适应产品重命名或法律原因导致的术语更改。

  • ACRIN的内容Cube是什么,它如何帮助分析已发布的内容?

    -内容Cube是ACRIN分析工具的一部分,能够扫描整个网页,帮助分析已发布内容的质量和一致性。

  • ACRIN如何帮助提高内容的可读性和用户效率?

    -ACRIN通过确保内容清晰、一致并符合企业标准,帮助提高内容的可读性,从而使用户能够更有效地使用内容。

  • ACRIN的生成性AI提供了哪些新功能?

    -ACRIN的生成性AI提供了复杂句子的重写建议,帮助作家简化难以理解的长句,提高内容的清晰度。

  • ACRIN的生成性AI如何处理数据保密性问题?

    -ACRIN使用私有基础设施处理数据,确保数据所有权清晰定义,保护数据不被用于其他目的。

  • ACRIN的生成性AI如何帮助改进复杂语言问题?

    -ACRIN的生成性AI可以接收复杂句子并提供更易读的替代版本,帮助作家改进内容的语言质量。

Outlines

00:00

🤖 AI助力内容质量提升

第一段主要介绍了AI助手如何帮助作家们提高内容质量。通过Acrin平台,作家可以在他们喜欢的应用中检查内容质量,并与企业写作标准进行比较,获得质量评分。Acrin会指出内容中的各种问题,如清晰度、术语一致性、包容性等,并提供改进建议。此外,AI助手还能根据企业标准生成内容建议,帮助作家快速生成符合标准的内容。

05:01

📝 内容创作流程与质量控制

第二段讨论了内容创作过程中的质量控制重要性。内容需要易于阅读,以便用户有效使用,避免可能造成损害或伤害的错误。作者通过Acrin平台的侧边栏工具直接在工作场所纠正形式问题,如拼写、语法和术语。此外,还有工具可以自动评分,帮助写作经理决定内容是否足够好以发布。所有这些流程最终汇集到分析数据库中,帮助改进整体内容质量。

10:02

🛠️ 集成AI的内容治理平台演示

第三段是关于Acrin作为一个AI驱动的内容治理平台的演示。演示者Mel CL和她的同事Ral展示了Acrin如何集成到内容创作过程中。Mel CL介绍了她作为高级技术客户经理的背景,而Ral则展示了Acrin在Adobe Experience Manager Guides中的集成。他们讨论了技术术语的一致性、内容的易读性以及如何使用Acrin来确保内容符合企业标准。

15:03

🔍 内容审核与质量提升实操

第四段展示了如何在Adobe Experience Manager Guides中使用Acrin进行内容审核和质量提升。演示者展示了如何检查和评分现有文档,并提供了详细的指导,帮助作者和编辑者改进内容。通过Acrin的侧边栏,作者可以直接在编辑器中接收实时反馈,并根据建议进行更改,以提高内容的清晰度和一致性。

20:04

📈 内容质量的持续监控与优化

第五段继续展示了内容质量的持续监控和优化过程。演示者介绍了如何在Acrin中配置不同的目标,以优化内容以适应特定的受众或产品。他们还讨论了如何使用Acrin来适应外部世界的变化,例如更新术语或适应新的法规要求,并确保所有发布的内容都符合企业的内容风格指南。

25:08

🔗 术语管理与内容一致性

第六段聚焦于术语管理,这是确保内容一致性的关键部分。演示者展示了如何在Acrin中管理术语,包括如何将术语分组、标记首选术语以及如何处理不适当的术语。他们还演示了如何更新术语,并在整个文档中应用这些更改,以确保内容与最新的企业标准保持一致。

30:11

🌐 内容发布后的监控与改进

第七段讨论了内容发布后的监控和改进。Acrin的内容立方体工具可以扫描整个公共网页,帮助企业识别和改进不符合内容风格指南的在线内容。演示者展示了如何使用内容立方体来识别需要改进的方面,如包容性语言,并提供了如何使用Acrin来提高内容质量的具体例子。

35:12

📝 利用生成性AI改进写作

第八段介绍了Acrin如何利用生成性AI来改进写作。演示者展示了Acrin的新功能,即通过大型语言模型提供复杂句子的重写建议。这将帮助作家更容易地解决难以手动修正的复杂语言问题,从而提高内容的清晰度和质量。

Mindmap

Keywords

💡AI助理

AI助理指的是利用人工智能技术来辅助人们完成各种任务的智能系统。在视频中,AI助理被用来帮助写作者提高内容质量,生成符合企业标准的新内容。例如,当写作者需要包含最佳实践时,AI助理能够提供内容营销协会的建议。

💡内容质量

内容质量是指所创作内容的优劣程度,包括清晰度、一致性、正确性等多个维度。视频中提到,Acrin通过比较写作者的内容与企业建立的写作标准来提供质量评分,并指出需要改进的地方。

💡术语一致性

术语一致性指在技术文档或产品内容中使用统一的技术术语,确保不同文档或组件描述中使用相同的词汇。视频中提到,Acrin确保所有话题使用相同的技术术语,以避免混淆。

💡写作标准

写作标准是企业或组织为保证内容质量和一致性而设定的一系列规范。视频中提到,Acrin生成的建议与企业写作标准对齐,帮助提高内容质量。

💡内容生成

内容生成是指利用技术自动创建文本、图像或其他形式的内容。视频中展示了AI助理如何帮助写作者生成新内容,例如询问内容营销协会的最佳实践。

💡写作效率

写作效率指写作者在创作内容时的速度和质量。视频中提到,通过Acrin的AI生成的建议,写作者能够快速获得反馈并改进内容,从而提高写作效率。

💡内容审核

内容审核是检查和评估内容以确保其符合特定标准和质量要求的过程。视频中提到,Acrin通过集成到写作工具中,允许写作者在创作过程中直接纠正形式问题,如拼写、语法等。

💡风险缓解

风险缓解是指采取措施减少或消除潜在风险的过程。视频中提到,通过使用Acrin,企业可以控制内容创作过程,确保内容的质量和一致性,从而降低风险。

💡内容管理

内容管理涉及创建、发布和维护数字内容的过程。视频中展示了Acrin如何帮助管理内容,包括在Adobe Experience Manager Guides中的集成和内容质量的监控。

💡生成式AI

生成式AI是一种能够生成新内容的人工智能技术,如文本、音乐或图像。视频中提到,Acrin正在开发利用生成式AI来提供复杂句子的重写建议,帮助写作者改进内容。

Highlights

Acrin为作家提供了企业级批准的生成性AI能力,以检查内容质量。

Acrin链接将内容与企业写作标准比较,并提供质量评分。

Acrin在侧边栏展示问题,便于作家点击修正。

Acrin提供AI生成的内容改进建议,与企业写作标准一致。

Acrin反馈涵盖清晰度、术语、包容性、正确性、可扫描性、一致性和风格等多个写作标准。

超过87%的内容问题提供可点击的指导,极大提高写作效率。

Acrin AI助手帮助生成新内容,并与内容营销研究所等资源进行整合。

Acrin AI助手能够根据企业写作标准生成并评分内容。

作家可以通过Acrin AI助手定制内容,如摘要。

Acrin集成到Adobe Experience Manager Guides中,提高内容管理效率。

Acrin通过侧边栏直接集成到工作场所,允许作者直接纠正形式问题。

Acrin的分析数据库帮助企业做出关于内容整体质量的决策。

Acrin通过定义术语和消息,以及添加质量门来控制内容创建过程。

Acrin通过实施风格指南来帮助组织中的所有作家保持内容一致性。

Acrin内容立方体能够扫描整个公开网页,识别与企业内容风格指南不一致的内容。

Acrin的包容性语言检查帮助避免对人们可能有害的语言使用。

Acrin的生成性AI能力提供了复杂句子的重写建议,简化了作家的修正工作。

Acrin使用私有基础设施处理数据,确保数据的保密性和所有权。

Acrin的编辑工具允许作家直接替换AI模型提供的句子建议。

Transcripts

play00:13

acrin get suggestions in AI assistant

play00:15

gives your writers access to Enterprise

play00:18

approved generative AI capabilities

play00:20

let's see how it works first writers use

play00:23

acrin to check content quality in their

play00:25

favorite authoring

play00:27

applications acry links will then

play00:29

compare your written content against

play00:31

your enterprises's established writing

play00:32

standards and provide a quality score

play00:35

issues are presented in the sidebar

play00:37

where writers can click to fix alignment

play00:39

issues across the content more complex

play00:42

issues like Clarity were often

play00:44

challenging to correct requiring more

play00:46

effort from the writer acrolink get

play00:49

suggestions provides AI generated

play00:51

content improvements

play00:58

[Music]

play01:00

acry links generate suggestions that are

play01:02

aligned with your Enterprise writing

play01:04

standards which improves content quality

play01:08

fast writers receive feedback on several

play01:11

writing standards such as Clarity

play01:13

terminology inclusivity correctness

play01:16

scannability consistency and

play01:19

style over 87% of content issues offer

play01:23

clickable guidance which drastically

play01:25

increases writer efficiency writers are

play01:27

able to check content as much as they

play01:29

like to make sure it meets your

play01:31

Enterprise standards now let's see the

play01:34

acinx AI assistant when writers need

play01:37

help generating new content they can

play01:39

launch the AI assistant from their

play01:41

favorite applications for Enterprise

play01:42

approved content generation let's say we

play01:45

want to include best practices in our

play01:47

new content we can ask what does the

play01:50

content marketing Institute say are best

play01:53

practices the ACR links AI assistant

play01:56

will research your request and provide

play01:58

you with answers

play01:59

[Music]

play02:08

[Music]

play02:14

acry linkx generates content that aligns

play02:16

with your writing standards and scores

play02:18

the

play02:19

content writers can quickly copy the

play02:22

generated

play02:24

content open the content in a new AI

play02:27

assistant window for comparison or

play02:29

bookmark the content so they can find it

play02:31

fast later writers can prompt the AI

play02:34

assistant to better customize the

play02:35

content for your writing needs such as

play02:37

summarization in this

play02:39

[Music]

play02:44

example then it's easy enough to copy

play02:47

and paste it into your

play02:49

[Music]

play02:54

documents to learn more about acry links

play02:57

visit ACR links.com thanks for having us

play03:00

at D World my name is Mel CL from arol

play03:03

links and my colleague R and I we are

play03:06

going to show you how arol link

play03:09

integrates into dat content creation

play03:11

process

play03:17

today so my name is m CL I am the senior

play03:20

technical account manager at arol links

play03:23

I normally work with our existing

play03:25

customers to help them integrate arol

play03:27

links into their processes and into

play03:29

their their technical organization

play03:32

basically I've been working technical

play03:35

documentation for over 20 years and have

play03:38

been yeah with different vendors in this

play03:41

field my colleague Al kunan is going to

play03:43

join later to show you arings in uh am

play03:49

guides he's the technical partner

play03:51

Solutions consultant at

play03:53

arings and he's been also a very long

play03:55

time in Pre of arings

play04:00

the speciality is that we want to show

play04:02

you how the AI powered content

play04:04

governance platform alings works

play04:06

together with your data

play04:08

content and as you all know data is

play04:13

topic oriented and topics uh are

play04:17

supposed to be self-contained but in the

play04:19

output they appear next to each other so

play04:23

that's then as some implications so

play04:25

shouldn't all the topics that you write

play04:28

use the same technical term terms for

play04:31

components for products Etc that's

play04:34

called

play04:35

terminology shouldn't they all use the

play04:37

same technical conventions something

play04:40

like unit and measures like how you uh

play04:44

write a work procedure don't use passive

play04:47

and work procedures for instance

play04:49

shouldn't they all use the same tone

play04:51

some organizations want to be rather

play04:54

friendly other want to be rather formal

play04:56

and short in technical documentation

play04:59

that's taste and there's reasons for

play05:01

that but okay let's see and after all

play05:05

the content created needs to be easy to

play05:08

read because only if the content is

play05:12

clear in writing your users will be able

play05:14

to use your content efficiently and not

play05:16

make mistakes that may lead lead to

play05:19

damage or even somebody being

play05:25

hurt we basically do that by having

play05:28

different steps in the authoring

play05:30

process we have a the so-called sidebar

play05:34

that integrates directly into the

play05:37

workplace am guides in this place but

play05:40

also in let's say Adobe framemaker Adobe

play05:42

in design and other tools that allows

play05:45

authors to write uh their content and

play05:50

directly correct all the formal issues

play05:53

like spelling grammar grammar uh style

play05:57

Etc

play05:58

terminology we also have tools that uh

play06:02

tag into the process so when something

play06:06

is signed off it'll be automatically

play06:08

scored by our coings and then the

play06:11

writing manager can make a decision is

play06:13

this good enough to be released or not

play06:15

and we'll show that later also and then

play06:19

uh this all flows into an analytics

play06:21

database and that then helps you make

play06:24

decisions about the overall quality of

play06:26

your content but also gives you input

play06:30

into uh opportunities to improve ours

play06:32

even further like adding more

play06:35

terminology perhaps activating

play06:37

deactivating CC guidelines so this would

play06:40

then lead to an improved cycle uh this

play06:43

is not something you do every day but uh

play06:45

we certainly suggest that you review

play06:47

your aring setup let's say every couple

play06:50

of

play06:55

months there's normally many different

play06:58

people writing product content is not

play07:01

only the technical documentation people

play07:04

that write product content there may be

play07:06

support writers or support Personnel Etc

play07:10

they all write content and it's not that

play07:13

you alone on in the world of content

play07:15

creation all those kind

play07:18

of uh teams would best be using

play07:20

something that like arings to control

play07:22

their uh copy what they send out to the

play07:25

end

play07:26

customers and with arol links you can

play07:28

basically have to assist assistance in

play07:31

all those

play07:32

process and we can also have something

play07:35

like a traffic light in the process as

play07:37

I've been been uh indicating before that

play07:41

certain content cannot proceed to The

play07:44

Next Step until it has reached a certain

play07:46

quality

play07:47

level that's probably accepted in the

play07:50

tech do Arena that those kind of

play07:52

processes exist if you have business

play07:55

writers that just write occasionally or

play07:58

need to shoot out content quickly like

play08:00

support people the the kind of depth of

play08:03

the process may be limited as we see it

play08:06

here in this

play08:07

diagram what are the effects of using

play08:10

arings in all of those

play08:13

processes the efficiency of the C

play08:15

creation process will be increasing

play08:19

without uh quality U uh

play08:22

sacrifices so if for instance the

play08:25

original author corrects and can correct

play08:29

most of formal issues directly then the

play08:33

reviewer can concentrate on subject

play08:35

matter correctness and there will be

play08:37

less rounds of review like Ping Pong uh

play08:40

issues be being sent from the writer to

play08:43

the reviewer reviewer sens back Etc so

play08:47

that should re reduce the number of

play08:48

review rounds and therefore make your uh

play08:51

entire content creation process much

play08:53

more

play08:56

efficient you also can mitigate risk and

play08:58

maintain control of the overall creation

play09:01

process so in aing you can Define your

play09:04

terminology and

play09:06

messaging and you can then add quality

play09:08

gates to your process either uh

play09:11

automatic as we show uh talked about

play09:13

earlier or manual that there are some

play09:16

reviews in the

play09:19

process and this will lead to less uh

play09:22

Techni less uh language issues in your

play09:25

content so terminology will be adhered

play09:27

to Etc and that will then lead to better

play09:29

content easier to read content less

play09:32

having less U technical issues

play09:36

also and uh many organizations have

play09:39

style guides uh guidelines how they want

play09:42

to write but if that's just a piece of

play09:46

PDF or a piece of paper printed out and

play09:48

sitting on the desk that doesn't help

play09:50

providers really to adhere to that

play09:53

arings is like an implementation of the

play09:55

sty guides if it go if it comes to

play09:57

linguistic issues and it gives then the

play10:00

tools to all the writers in your

play10:02

organization to really write content

play10:04

that's in line with the uh procedures

play10:08

and and with the tone that you want to

play10:13

produce we now show you how that is

play10:16

being implemented in uh Adobe experience

play10:19

manager guides and that's something that

play10:21

my colleague Ral will do in a moment so

play10:25

uh we have as we said earlier the

play10:27

sidebar the directly inte integrates

play10:29

into the arol links process and we also

play10:33

have a uh the arink course directly

play10:36

attached to the topics so that you have

play10:39

also a record of the

play10:41

quality and you also have tools to then

play10:43

see post publishing how the quality also

play10:46

on there hello everybody I'm and today I

play10:50

like to show you how arings integrates

play10:52

into am

play10:54

guide so most of our customers are not

play10:57

starting from scratch so they are

play10:59

already have a lot of documents in Adobe

play11:02

am guides um a lot of deta Doc documents

play11:05

with several topics and we generally

play11:09

recommend to first get an overview of

play11:12

all your content so that you know where

play11:14

to focus with your editing

play11:17

capacities so um I set up here a sample

play11:21

project that you see here um it's a

play11:24

project about or a d map about um an oil

play11:28

change on a car and um so let's get an

play11:33

over first so I switch here to the Adobe

play11:38

um assets

play11:39

UI and you see all the content here

play11:42

again I select all the files that are in

play11:46

in that folder or in that uh yeah and uh

play11:50

click check for um the selected pages

play11:53

with EOL links um here in the toolbar so

play11:56

the content now gets checked and scored

play11:59

you see here is uh the score column um

play12:03

if I refresh you see that the status

play12:06

changed to pening and um if I refresh

play12:09

further you see that the check results

play12:11

are um getting shown here and um so now

play12:17

all content is scored um you see um some

play12:21

yellow here you see also green you see

play12:23

also number here and so the aing score

play12:26

goes uh from 0 to 100

play12:30

where um the lower part is red like a

play12:33

traffic light um yellow for Content that

play12:36

is starting to get okayish and green for

play12:39

Content that is okay so um green starts

play12:42

at 80 and um we usually consider um

play12:46

content that is 880 um as good to go but

play12:50

of course you can Define your own

play12:52

standards and you can also um uh tweak

play12:56

your uh digitalization of your content

play12:59

style guide regarding your target market

play13:02

or your target personas um we will see

play13:05

how that works later and you can have uh

play13:09

several um different targets that you

play13:12

optimize your content for um so that

play13:14

you're able to be very specific for

play13:18

certain audience or certain um yeah uh

play13:22

product or or something like that so um

play13:27

looking at all these parts where to

play13:28

start of course start with um the

play13:31

content that needs the most Improvement

play13:33

and in that case um we look at uh this

play13:38

document here um it's feasible to pull

play13:41

up an actual scorecard that's available

play13:43

in all of our Integrations um it uh

play13:46

directly shows an an overview uh over

play13:50

your content it um shows all the issues

play13:53

that acons found with the different gos

play13:57

um like uh clarity like consistency

play14:00

correctness inclusive language tone

play14:02

words and phrases and uh all other goals

play14:06

that you configure for that specific um

play14:08

style guide and um so you already get a

play14:13

good overview here and uh get also

play14:16

recommendations uh what you should uh

play14:19

change um but uh of course you want to

play14:24

improve your content so we switch to the

play14:28

addit mode here in uh am

play14:32

guides and um so uh you see it's loaded

play14:36

now in into the editor I pull up here in

play14:40

um in the toolbar the acing sidebar and

play14:44

check so that I get um Hands-On guidance

play14:47

and how to improve and um for example

play14:52

acing found here a contraction and

play14:56

that's really kind of how I configured

play14:58

the

play14:59

um digitalization of a uh demo style

play15:02

guide in that case for for that so I um

play15:06

configured it in a way that I don't want

play15:08

to have uh contradictions of course you

play15:11

can also configure it the opposite way

play15:13

around actl is there for making your

play15:15

content consistent and being really

play15:18

aligned to how you want to write so um

play15:22

here you see uh the card that gives uh

play15:24

to the writers and editors uh handson

play15:27

guidance um what what to avoid you see

play15:31

also um why that's the case uh you also

play15:35

can click here on further information to

play15:37

get a really detailed um description is

play15:40

that is not sufficient here um and in a

play15:44

lot of cases you also get a uh

play15:46

suggestion that you can apply with one

play15:49

click so I'm going to do that now and

play15:51

click here you see it got automatically

play15:53

replaced we automatically jump to the

play15:55

next issue that is still um uh in place

play16:00

and uh you see also here's a score um

play16:03

that you've seen before I going to check

play16:08

again and you see it's already improved

play16:11

but um there are still issues in and I

play16:15

uh going to correct uh that one as well

play16:17

so you see here there is an issue that

play16:21

um either this word should be um plural

play16:25

or that uh word needs to be in uh kind

play16:29

of the uh uh one that is uh for singular

play16:33

work so you see AC provide both variant

play16:36

of that so that I can uh now decide that

play16:39

it's one vehicle which sit there I can

play16:43

check again and um see that the content

play16:46

get

play16:48

approved so now um that I changed the

play16:52

content I can uh save the content of

play16:54

course there are more issues powered

play16:56

here you see the score approved to 88 um

play16:59

and uh there are more uh more detailed

play17:03

feedback here like uh this um uh

play17:06

paragraph is too complex so uh I get a

play17:10

lot of guidance how I can improve it by

play17:12

for example splitting up the

play17:14

sentences um having uh too many uh

play17:19

prepositions in in that sentence uh and

play17:21

so on and so on so there are different

play17:24

different aspect that all would uh

play17:26

simplify the content for the readers so

play17:29

that your content gets understood by

play17:31

your

play17:32

readers um for now I leave it like that

play17:36

uh of course you can can or in the

play17:38

production environment you should uh

play17:40

focus on on a bit more um but for the

play17:43

demo I keep it to that I save the file I

play17:46

switch back to my um overview here and I

play17:51

also score here

play17:53

again and we now see that uh the score

play17:57

gets updated here as well and see that

play17:59

for my interre

play18:00

publication my uh content is in a good

play18:03

shape and I'm ready to good and good for

play18:06

go for

play18:09

publication so um if you look at

play18:13

the um content supply chain then kind of

play18:18

from um authoring it's covered by

play18:21

pulling up the sidebar checking your

play18:23

content while you're authoring um if you

play18:26

look at uh the quality um before

play18:29

publishing we can um use that as quality

play18:32

gate so before I um hit the publish

play18:36

button I check all my content and ensure

play18:39

that the that it's good to go and that

play18:41

no terms are in that I don't want to

play18:44

have in but of course if you look at the

play18:47

content supply chain um then um you

play18:51

might want to do uh more like your

play18:54

content um might be in a good shape

play18:58

while while you published it but the

play19:00

outside world change okay let's um have

play19:03

a look how it looks if the outside world

play19:06

changes so for example um you see uh

play19:11

here um our digitalization of um the um

play19:16

style guide um for a certain Target um

play19:20

so in the ACT UI it's called Target and

play19:24

um you can configure very fine granular

play19:28

how you want to um uh align your content

play19:34

uh with with your your goals so for

play19:36

example you can configure how long your

play19:38

sentence should should be in different

play19:41

context like if it's in a list item or

play19:44

in a normal sentence or in a title um

play19:47

about uh for example uh which difficult

play19:51

words you want to avoid or where to

play19:53

allow you can also be very context

play19:55

specific so to allow certain certain

play19:58

things only in certain Fields so for

play20:00

example if you have a uh field that is

play20:03

for um further information or something

play20:06

you can um make your your language uh

play20:10

more complex there or if you look at uh

play20:14

very easy steps you can also configure a

play20:17

certain uh context to be more strict um

play20:20

than than others and uh you can also or

play20:25

you can also configure here the

play20:27

difficulty of sentences um in range from

play20:30

very simple to complex um and a lot more

play20:34

phenomenas um and that's something where

play20:37

aing comes with uh good standards a good

play20:40

uh something we we recommend but it's

play20:43

also something that uh during onboarding

play20:46

with arings um our uh Services teams

play20:49

will work with you to uh help you

play20:53

digitalize your content style guide also

play20:57

um a couple of our rules or a lot of our

play21:00

rules also have the feature to be um

play21:04

consistent for example with in uh your

play21:08

your document so you can uh for example

play21:11

either select um for a title um AP style

play21:15

or style or you can decide to just be

play21:20

consistent within in the document um of

play21:23

course if you're in Deer environment as

play21:25

you're usually working on a on a topic

play21:27

level uh it most makes sense to to have

play21:31

one style that you apply for all of your

play21:34

detail content so that that you're free

play21:37

in assembling and disassembling your

play21:40

your content and putting it together how

play21:42

your needs of your publication

play21:44

is so you see also correctness inclusive

play21:48

language and uh couple more in my sample

play21:52

I going to add now a uh certain domain

play21:56

um so I pre-prepared um a domain uh

play22:02

called Automotive new so um pretending

play22:06

that there's new um terminology for uh

play22:09

for our use case and uh now akings asked

play22:13

me for uh reloading uh the terminology

play22:16

so I agreed to that and um I going to

play22:19

switch back now here to the terminology

play22:22

management of act Lings um that the UI

play22:26

where can uh manage um actl to

play22:28

technology and um I created some terms

play22:32

here now so I said that automobile car

play22:35

and vehicle belong together so they are

play22:38

grouped see here uh this check marks

play22:41

that they're grouped together those uh

play22:43

line here represents that they belong

play22:45

together and um automobile is unsuitable

play22:49

car is preferred term and vehicle is

play22:52

another unsuitable term and um that's

play22:55

really kind of what could happen to you

play22:57

um or with the Daily Business like um

play23:00

someone in the company decides that a

play23:02

certain um product gets renamed that the

play23:05

company gets renamed that uh you're not

play23:08

allowed to call uh a certain word or

play23:11

certain thing like a certain word

play23:13

anymore because du to legal reasons um

play23:16

or other things maybe you realize that

play23:19

um there's a lot of confusion for

play23:21

example here oil filter R wrange is uh I

play23:25

marked as the unsuitable and call The

play23:27

Tool uh 7 831 just a random name but uh

play23:31

to represent that uh if you have for

play23:34

example customer a lot of customers cor

play23:36

question what exactly that is you might

play23:38

be able to specify it very very uh

play23:41

detailed so that um that uh there's a

play23:46

lot of clarity in in what you're right

play23:49

um acing also like you see in the

play23:51

brackets here uh is capable of um uh

play23:56

checking for flacs other words so if you

play23:58

would uh you would only need to add car

play24:01

for it but it would also uh work with

play24:05

cars so if there's cars um it would also

play24:08

apply our suggest vehicles in that case

play24:12

so that's uh also very uh yeah powerful

play24:16

way that saves you a lot of time when

play24:18

dealing with uh uh with

play24:22

terminology so going back to to the

play24:24

content I now uh select all of the stuff

play24:28

here and I hit uh the eons button and I

play24:34

uh going to refresh um the page and you

play24:37

see that uh scores not updated and you

play24:40

now also see that um a lot of those

play24:43

topics that had been green before are um

play24:46

now in a yellow State and I going to um

play24:51

edit it uh again to see what what really

play24:56

um really uh changed or what uh what

play24:59

issues are appearing now and um you see

play25:03

now that uh for example uh vle got uh

play25:07

flagged and uh you also see that

play25:10

vehicles in that case it's also phone um

play25:14

and uh you also see that it uh Flags

play25:17

again here um I'm able to tell applinks

play25:21

to replace all recurrentes with one

play25:23

click um so once I click here um vehicle

play25:28

is replaced by car and in all

play25:31

occurrences of that uh topic so um that

play25:35

makes it very handy to align the uh the

play25:39

content to the updated uh

play25:43

terminology and uh same for oil filter

play25:46

Ranch I'm able to uh apply the new um

play25:51

suggestion here I check again and I see

play25:54

that the score improved again but oh

play25:57

AR's found another issue now it's um a

play26:00

tool and not n tool so I um fix that as

play26:05

well and uh now I'm good to go with my

play26:09

content again uh so I going to save that

play26:12

and of course in uh real word example

play26:16

you would go back and also apply these

play26:18

corrections to all other content that

play26:21

needs Corrections the last part of our

play26:23

demo is not really inside of aim guides

play26:27

but beyond so what we've seen so far is

play26:30

inside of am guides and um these are the

play26:33

data topics that you're really editing

play26:35

and uh where you have direct influence

play26:38

but once you publish things um topics

play26:41

get resolved um variables get resolved

play26:44

uh you might deploy to several different

play26:49

um yeah targets like uh content delivery

play26:52

platform like am like a web server

play26:56

whatever um so and and in those cases

play26:59

there might be a menu be generated U

play27:03

maybe there are JavaScript parts that

play27:05

get resolved later um maybe there's some

play27:07

legal statements that get added to all

play27:09

of your pages um there might be another

play27:12

header fooders that gets added to your

play27:14

pages and all these things could

play27:17

potentially not be aligned with your

play27:19

Enterprise content style guide that we

play27:21

checked so far with with it kind of the

play27:23

real content so um what aklin offers for

play27:27

that is is um AC content Cube and acin

play27:31

content cube is part of the AC analytics

play27:33

offering it's uh capable of scanning an

play27:36

entire public web

play27:38

page and uh you see for example here um

play27:43

the ACX blog post that I that I checked

play27:47

and uh I select the eons block post

play27:49

because there are couple of special

play27:51

things with that so it contains um for

play27:55

example webinars where we do have uh

play27:57

transcript of that that we cannot change

play28:00

um so people set things and they're

play28:02

they're there and even if they don't

play28:04

align with our content sty guide anymore

play28:06

uh we cannot change it um there are uh

play28:10

blog posts about uh certain phenomenas

play28:13

and we need to phrase those phenomenas

play28:16

otherwise um we cannot talk about them

play28:19

so we need to keep them on on that page

play28:22

and so it's a very good uh place to um

play28:25

kind of demonstrate how you get inside

play28:28

into your content um of course you would

play28:30

apply that to your um technical content

play28:34

that uh you published and uh that helps

play28:37

you really uh yeah dive into your

play28:40

content and identify um the

play28:44

um content on the documents that uh you

play28:48

need to focus on um a lot of our

play28:51

customers um start focusing on a certain

play28:54

aspect um so they might have an

play28:56

inclusivity initiative

play28:58

and um so uh you see here the the

play29:02

different gos and um I like to uh use

play29:06

intrusive language in that case uh to

play29:08

demonstrate uh what you see here is um

play29:12

the acon score um so you see all the

play29:15

content is either green or um or Yello

play29:18

is uh on on that blog post so it's not

play29:21

as bad um but if we look at certain

play29:25

aspects for example inclusive language

play29:27

then you seees that there are a couple

play29:29

of red um yeah documents and uh I still

play29:34

would have uh tough time to get all

play29:37

through those uh documents and I

play29:40

wouldn't like to have more guidance

play29:42

where to start so and that's where um

play29:44

actual performance data come into the

play29:46

picture so um content cube is able to

play29:50

connect to um analytics platforms uh for

play29:53

example Adobe analytics and um so I can

play29:56

pull in for example the

play29:58

page views and that's what I going to do

play30:01

here and now you see okay uh the good

play30:03

thing the um documents that get accessed

play30:07

frequently they are green but uh if you

play30:10

look here there's a red one that uh is

play30:13

bigger than than others and uh so that's

play30:16

for inclusive language and Page shits so

play30:19

I select that as um um candidate or uh

play30:23

content that I would like to to improve

play30:26

so I can uh just into that I see here

play30:29

also the inclusive language is not as

play30:31

bad um so we set very uh high goals here

play30:35

so you see even with the U 90 uh3 uh

play30:39

it's it's sh red here um that's

play30:42

configurable for um for uh that view um

play30:47

and uh you also see here with all

play30:50

aspects of the content how it performs

play30:53

so ideally of course it would be all

play30:55

blue in that case uh but you see for

play30:58

example there's Clarity bit correctness

play31:00

a bit and tone and um so but in in that

play31:04

Dem I like to uh focus on inclusive uh

play31:08

language and uh you can see here for

play31:11

example um directly that um what are the

play31:15

issues and uh you see for example uh

play31:18

that that phrase was used and um in

play31:21

general should be used uh that phrase or

play31:24

that suggestion that ACS provide you can

play31:27

also get further guidance on these uh

play31:29

guidelines um within acing and uh you

play31:33

also have a link here to um to jump to

play31:36

the actual content that is live out

play31:38

there so that you see how the issues are

play31:41

in in the real world and uh here you

play31:44

already see kind of that um the content

play31:47

uh that uh got uh flaged here really

play31:51

talks about the issues and how you

play31:53

should avoid them and what you should do

play31:57

uh instead so uh that um yeah that's

play32:01

also the reason why these inclusive

play32:03

language issues are out there on the A's

play32:06

website of course we're actually

play32:07

addressing them and talking about them

play32:09

um how to uh avoid them and how to

play32:12

improve the content of your document and

play32:15

that's also something that um people

play32:17

would learn out of the usage of applinks

play32:20

um by getting trained through the

play32:23

sidebar um and improv their own writing

play32:27

um looking at at uh that dator um

play32:29

there's also a lot more that you can do

play32:32

out of that so if you think about all AI

play32:35

topics um that are out there where you

play32:38

want to use your content for any AI

play32:41

purpose like training an llm like using

play32:44

that content to automatically answer um

play32:46

customer questions you really want to

play32:49

avoid that bad content goes into your um

play32:53

AI system and is repeated and shown to

play32:57

the users again so especially if you

play33:00

think if you think about inclusive

play33:02

language that would be probably

play33:04

something that that is really could be

play33:06

very harmful for for people and so you

play33:09

you want to strictly avoid that but you

play33:11

also want to make sure that um your AI

play33:16

um provides output that is aligned with

play33:19

your content style guide and that's

play33:21

where where act links helps by

play33:23

identifying that content and you can use

play33:26

the aclink score as indicator for um

play33:29

good content or content that you need to

play33:31

improve and um yeah there's also another

play33:35

AI offering that um akings has um on

play33:40

current development and that's something

play33:42

um Michael will talk uh about in a

play33:45

minute thank you very much back to you

play33:48

Michael thanks R for showing us our

play33:51

links in uh am guides there's something

play33:55

that we've just been adding to our

play33:57

products well the process of adding to

play33:59

our product at the very moment and that

play34:01

I would like to explain last in our

play34:05

presentation so uh that's basically

play34:08

generative Ai and arol links arol has

play34:11

always been an AI system we've been

play34:14

around natural language processing for

play34:16

22 years we've been adding machine

play34:18

learning as an addition some years ago

play34:21

but now generative AI offers us uh

play34:25

additional possibilities that we are

play34:26

going to use so we can with jange of AI

play34:30

offer uh rewrites of complex sentences

play34:34

or suggestions how to change those

play34:36

sentences and that's a going to be a

play34:39

major Game Changer I think for many

play34:41

writers since those are very difficult

play34:43

to fix for them generative AI is being

play34:47

talked about a lot but many people have

play34:50

concerns and that's about the

play34:52

confidentiality of the data being

play34:54

processed uh we are uh using a private

play34:58

infrastructure so this is not just um

play35:02

some some public tool uh where your data

play35:04

is being being used for training of of

play35:07

the generic models for instance the data

play35:09

ownership is therefore clearly defined

play35:12

it's your data not somebody else's data

play35:15

and we're using a future proof

play35:16

infrastructure that will allow us also

play35:19

to grow this service over

play35:22

time let me give you a short demo of

play35:24

that so that's another edit tool that's

play35:28

well known you see basically the sidebar

play35:31

it looks quite similar here compared to

play35:33

the one that you saw in the tool that

play35:36

ral's been

play35:38

showing let me make this a bit larger

play35:41

perhaps and there are the sidebar cards

play35:45

that you also saw before what we're

play35:47

going to addit add now is basically this

play35:49

guest suggestions button this will uh

play35:53

take the sentence that's been been

play35:55

highlighted send it over to to the large

play35:58

language model and ask for a

play36:01

rewrite the large language model uh uh

play36:05

then uh sends back a sentence that's in

play36:07

line with all the guidelines defined for

play36:09

this target group and the writer can

play36:12

directly replace the sentence from

play36:16

here there's another one here so AR has

play36:19

found out that this sentence over forign

play36:22

bit lines is difficult to

play36:24

read and uh that's what we all have been

play36:27

doing for years it's difficult to read

play36:29

because it's a long sentence because

play36:31

there are propositions because there's a

play36:34

parenthesis inside and there's a noun

play36:36

cluster also but now the writer would

play36:39

have had to rewrite this sentence into

play36:41

something more readable and that's

play36:42

something that very difficult to do and

play36:45

here again the AI model receives the

play36:49

sentence and it then sends back uh a

play36:53

suggestion on how to replace this you

play36:56

can also ask for suggestion if you like

play36:59

so if you don't like the sentence here

play37:01

so this is something that you can then

play37:02

also iterate over and and see a couple

play37:05

of suggestions of the

play37:08

system we think that this is a major uh

play37:11

Improvement of the system because now uh

play37:15

fixing those complex linguistic things

play37:18

are quite easy to

play37:19

do thanks for being with us today I hope

play37:23

that uh we could share some insight into

play37:26

arings and how to use it best

play37:29

uh if you have any questions don't

play37:31

hesitate to contact Ral or myself you

play37:34

see our email address here or visit our

play37:36

corporate website for more

play37:38

information thanks a lot

play37:54

[Music]

Rate This

5.0 / 5 (0 votes)

Étiquettes Connexes
AI助手内容治理写作标准企业效率技术文档自然语言处理机器学习内容质量作者工具术语管理
Besoin d'un résumé en anglais ?