You see entropy is information leaking. When it comes to passwords (and secure systems in general) you want to leak as little information as possible. Otherwise an attacker has information they can use to their advantage.

Download the billions of breached passwords and blacklist them all. Attackers have a copy; so should you.

This isn’t the first time Kerckhoffs’ Principle has come up. I specifically discussed it when talking about creating good, strong Master Passwords, when I said that we should use a system for coming up with Master Passwords that doesn’t lose its strength if the attacker knows the system that we used

Kerckhoffs’ Principle states that you should assume that your adversary knows as much about the system you use as you do. This is why – despite what I may have said on April Fools Day last year – security experts are skeptical of security systems that hide the details of how they operate. They are particularly skeptical of systems that derive their security from keeping the details of how they work secret. I could go on at great length about why openness about the system improves security. Indeed, my first draft of this article did go on at great length.

"the enemy knows the system"

"one ought to design systems under the assumption that the enemy will immediately gain full familiarity with them"

School for many people is a place to get fed, a place to feel safe, a place to get encouraged. It’s a place to be around people who share your desire to learn. Now they are cut off from that, and some of that can’t be duplicated easily online.

Relational databases are designed around joins, and optimized to do them well. Unless you have a good reason not to use a normalized design, use a normalised design. jsonb and things like hstore are good for when you can't use a normalized data model, such as when the data model changes rapidly and is user defined. If you can model it relationally, model it relationally. If you can't, consider json etc.

Joins are not expensive. Who said it to you? As basically the whole concept of relational databases revolve around joins (from a practical point of view), these product are very good at joining. The normal way of thinking is starting with properly normalized structures and going into fancy denormalizations and similar stuff when the performance really needs it on the reading side. JSON(B) and hstore (and EAV) are good for data with unknown structure.

Data Model

A website called Open Food Facts, run by mostly French volunteers, has started the herculean labour of creating an open database of packaged foods around the world and listing where they fit into on the Nova system.

I use this to keep information about processes that were running at any time during last five days. These are 1-min snapshots so something might get lost but I think it is good enough for me. I want to have some data available when I discover there was a peak in resource usage (I use munin for that). I haven't found a better way to keep track of past processes (tried psacct).

Indexes are a cornerstone of information retrieval, and the basis for today's modern search engines.

This is known as collapsed forwarding.

four of the more important ones are caches, proxies, indexes and load balancers.

benchmark different alternatives, understand how the system will fail, and have a solid plan for when failure happens.

with the Flickr architecture each shard would need to be updated or searched (or a search service would need to be created to collate that metadata—which is in fact what they do).

Finally, this separates future concerns, which would make it easier to troubleshoot and scale a problem like slow reads.

Deconstructing a system into a set of complementary services decouples the operation of those pieces from one another. This abstraction helps establish clear relationships between the service, its underlying environment, and the consumers of that service. Creating these clear delineations can help isolate problems, but also allows each piece to scale independently of one another. This sort of service-oriented design for systems is very similar to object-oriented design for programming.

It's not for beginners, but it's extremely flexible once you get a handle on things. If sticking to a system is hard for you, I highly recommend giving this system a try.

Best to-do list app for taking control of your to-do list

todo.txt to track tasks

the system

it is certainly more creditable to cultivate the earth for the sustenance of man, than to be the confidant, and sometimes the accomplice, of his vices; which is v1_117the profession of a lawyer

Now using sudo to work around the root account is not only pointless, it's also dangerous: at first glance rsyncuser looks like an ordinary unprivileged account. But as I've already explained, it would be very easy for an attacker to gain full root access if he had already gained rsyncuser access. So essentially, you now have an additional root account that doesn't look like a root account at all, which is not a good thing.

It is possible to do a successful file system migration by using rsync as described in this article and updating the fstab and bootloader as described in Migrate installation to new hardware. This essentially provides a way to convert any root file system to another one.

rsync provides a way to do a copy of all data in a file system while preserving as much information as possible, including the file system metadata. It is a procedure of data cloning on a file system level where source and destination file systems do not need to be of the same type. It can be used for backing up, file system migration or data recovery.

I am familiar with using rsync to back up various files from my system but what is the best way to completely restore a machine.

I run the script daily, early every morning, as a cron job to ensure that I never forget to perform my backups.

There are many options for performing backups. Most Linux distributions are provided with one or more open source programs specially designed to perform backups. There are many commercial options available as well. But none of those directly met my needs so I decided to use basic Linux tools to do the job.

CloneZilla works perfectly. It produces small image files, has integrity check and works fast. If you want to use third device as image repository you should choose device-image when creating image of the first disk and then image-device when you restore it to second disk. If you want to use only two disks - you should use device-device mode. Optionally you may want generate new UUIDs, SSH-key (if SSH server installed), and change hostname.

While there are so many tools to backup your systems, I find this method super easy and convenient, at least to me. Also, this method is way better than disk cloning with dd command. Because It doesn’t matter if your hard drive is different size, or use different filesystem.

don't mix application dependencies with those of the build system

Can I ask people to upvote my submission? No. Users should vote for a story because they personally find it intellectually interesting, not because someone has content to promote.

Twitter offers two distinct benefits to engaging learners. First of all, it allows learners to respond to classroom discussions in a way that feels right for them, offering shy or introverted students a chance to participate in the class discussion without having to speak in a public forum. Secondly, it allows students to continue the conversation after class is completed, posting relevant links to course material, and reaching out to you (the educator) with additional thoughts or questions.

Early alignment is everything

While these are excellent frameworks for evaluating instructor-level change, our field is pivoting from an emphasis on 1:1 work or workshops to longer-term, systemic change initiatives

Sen. Bernie Sanders, I-Vt., with his “Medicare for All” legislation—suggest Americans would enjoy a health care utopia if only the government took over.

A “single-payer” health system is a government-controlled health care system. Government is the “single-payer.” In most versions of single-payer, most private health insurance is either outlawed or restricted, and most public health programs are absorbed into the single, national health insurance program.

 In Switzerland, which has a private universal health care system, only 25.9% waited at least 4 weeks.

third of Americans reported they didn’t go to the doctor

The United States is the wealthiest nation

by Blahous’s own estimates that single-payer would save Americans more than $2 trillion over a decade.

Funding for his plan is still being debated, but he has come up with multiple suggestions that could be passed.

Medicare eligibility age would be lowered to 55, second year to 45, third year to 35, and fourth year everyone would be covered.

You will not be able to keep the insurance you have

tudies show that costs would be between $25 Trillion and $32 Trillion over 10 years. YouGov/Economist Poll, April 2-4, 2017      $32 Trillion sounds like quite a high number.

Removing the current system would a gigantic task.

Health care spending was 12.4% of GDP in 2016. That is approximately $7,919.00 per person. There were 11.6% of people who skipped prescriptions because of cost.

Health care spending was 11% of GDP in 2016. Approximately $4,600.00 per person. 7.8% of patients skipped prescriptions because of cost. The life expectancy was 85.5 years in 2015.  

Health Care spending was 10.6% of Canada’s GDP in 2016 and 10.5% of patients skipped prescriptions because of cost.

Jillian Maynard, from the University of Hartford, and Jeremy Anderson, from Bay Path University, led a session on developing a strategic plan for OER on your home campus. I went to that one having previously selected a different one -- born to be wild, that’s me -- because upon closer reading, I made the connection to a law that NJ recently passed requiring public colleges to develop plans for OER (or “inclusive access”)

inferiority of your connexions? — to congratulate myself on the hope of relations, whose condition in life is so decidedly beneath my own?”

The goal here is explicitly not to improve the state of the art in the narrow domain of restaurantbooking, but to take a narrow domain where traditional handcrafted dialog systems are known toperform well, and use that to gauge the strengths and weaknesses of current end-to-end systemswith no domain knowledge

Unsurprisingly, perfectly coded rule-based systems can solve the simulated tasks T1-T5 perfectly,whereas our machine learning methods cannot. However, it is not easy to build an effective rule-based

SUPERVISEDEMBEDDINGMODELS

We implemented a rule-based system for this task in the followingway. We initialized a dialog state using the 3 relevant slots for this task: cuisine type, location andprice range. Then we analyzed the training data and wrote a series of rules that fire for triggers likeword matches, positions in the dialog, entity detections or dialog state, to output particular responses,API calls and/or update a dialog state. Responses are created by combining patterns extracted fromthe training set with entities detected in the previous turns or stored in the dialog state. Overall webuilt 28 rules and extracted 21 patterns. We optimized the choice of rules and their application priority(when needed) using the validation set, reaching a validation per-response accuracy of 40.7%. Wedid not build a rule-based system forConciergedata as it is even less constrained.

Finally, the policy network output is generated bya three-way matrix transformation,

a distributed representationgenerated by an intent network and a probabilitydistribution over slot-value pairs called the beliefstate

In learning such neural network based dialogmodel, we propose hybrid offline training and on-line interactive learning methods. We first let theagent to learn from human-human conversationswith offline supervised training. We then improvethe agent further by letting it to interact with usersand learn from user demonstrations and feedbackwith imitation and reinforcement learning.

We design neural net-work based dialog system that is able to ro-bustly track dialog state, interface with knowl-edge bases, and incorporate structured queryresults into system responses to successfullycomplete task-oriented dialog.

These system components areusually trained independently, and their optimiza-tion targets may not fully align with the overallsystem evaluation criteria (e.g. task success rateand user satisfaction). Moreover, errors made inthe upper stream modules of the pipeline propa-gate to downstream components and get amplified,making it hard to track the source of errors

or joint modeling of intent detection and slot filling, weadd an additional decoder for intent detection (or intent clas-sification) task that shares the same encoder with slot fillingdecoder.

The attentionmechanism later introduced in [12] enables the encoder-decodermodel to learn a soft alignment and to decode at the same time.

We present a general solution towards building task-orienteddialogue systems for online shopping, aiming to assist on-line customers in completing various purchase-related tasks,such as searching products and answering questions, in a nat-ural language conversation manner. As a pioneering work, weshow what & how existing natural language processing tech-niques, data resources, and crowdsourcing can be leveragedto build such task-oriented dialogue systems for E-commerceusage. To demonstrate its effectiveness, we integrate our sys-tem into a mobile online shopping application. To the bestof our knowledge, this is the first time that an dialogue sys-tem in Chinese is practically used in online shopping scenariowith millions of real consumers. Interesting and insightful ob-servations are shown in the experimental part, based on theanalysis of human-bot conversation log. Several current chal-lenges are also pointed out as our future directions

Main actions that areconsidered in the online shopping scenario include

To deal with the problem we mentioned, our work focuson using three kinds of data resources that are common tomost E-commerce web service provider or easily crawledfrom webs, including: (i) product knowledge base, which isprovided by the E-commerce partner and contains structuredproduct information; (ii) search log, which is closely linkedwith products, natural language queries and user selectionbehaviors (mouse click); (iii) community sites, where userpost their intents in natural language and can be used to minepurchase-related intents and paraphrases of product-relatedterms. Besides, we show that crowd sourcing is necessary tobuild such AI bot

CHATBOT: Architecture, Design,& Development

The first step is lexical analysis, i.e. word segmentation and part-of-speech (POS)tagging. The words and POS labels are used as features in the subsequent models. Forthe shared task we used HanLP [1] as our Chinese lexical analyzer.

Each rule is of the form “if thequery q is listed in a particular lexicon L, and the preceding queries and their predicteddomain labels satisfy certain conditions, then q is assigned a certain intent label and,with the exception of short commands, the entire q is regarded as a slot of the typecorresponding to L.” The rules are arranged in sequential order in accordance with theirpriorities

Figure 1 shows the framework of our SLU system, which consists of the context-dependent rules for entity-only queries and the context-independent model for querieswith IISPs. The entire system feeds the query to the rules first. If the rule-based compo‐nent returns null result, that means the query is judged to contain IISPs and the model-based component will continue to process it. Otherwise, it means the query is regardedas entity-only and the result of the rules is returned as the final output

s in real use cases of dialog systems, the queries in the shared task can be roughlydivided into two kinds, viz. queries with intent-indicating salient phrases and querieswithout. By intent-indicating salient phrase (IISP) it is meant a phrase in the query thatshows the intent of the query. E.g. the phrase “” in the query “” andthe phrases “” in the query “” are IISPs.

Retrieval-based MethodsRetrieval-based methods choose a response from candidateresponses. The key to retrieval-based methods is message-response matching. Matching algorithms have to overcomesemantic gaps between messages and responses [28].

TASK-ORIENTED DIALOGUESYSTEMSTask-oriented dialogue systems have been an important branchof spoken dialogue systems. In this section, we will reviewpipeline and end-to-end methods for task-oriented dialoguesystems.

2.2 End-to-End Methods

TASK-ORIENTED DIALOGUESYSTEMS

2.1.3 Policy learning

A statistical dialog system

Dialogue State Tracking

Slot filling

Language Understanding

setting up objectives

A composite process, remember, is organized from both human processes and computer processes

Spoken language understanding (SLU) comprises two tasks, intent identification andslot filling. That is, given the current query along with the previous queries in the samesession, an SLU system predicts the intent of the current query and also all slots (entitiesor labels) associated with the predicted intent. The significance of SLU lies in that eachtype of intent corresponds to a particular service API and the slots correspond to theparameters required by this API. SLU helps the dialog system to decide how to satisfythe user’s need by calling the right service with the right information

对话管理也可以看成是一个分类任务，即每个对话状态和一个合适的对话动作相对应．和其它有监督的学习任务一样，分类器可以从标注的语料库中训练得到．但是，在某状态下系统应该选择的动作不能仅仅是模仿在训练数据中同一状态对应的动作，而应该是选择合适的动作能够导致一个成功的对话．因此，把对话过程看成是一个决策过程更为合适，从而根据对话的整体成功来优化动作的选择过程［３２］．因而这是一个规划问题，并且可以用强化学习［３３］方法学习获得最优的结果

对话系统从本体构成和业务逻辑角度，可分为领域任务型和开放型的信息交互．领域任务型系统针对具体应用领域，具有比较清晰的业务语义单元的定义、本体结构以及用户目标范畴，例如航班查询、视频搜索、设备控制等等，这类交互往往是以完成特定的操作任务作为交互目标；而开放型信息交互则不针对特定领域，或说面向非常广泛的领域，交互目标并非业务任务，而是满足用户其它方面的需求，例如开放的百科问答、聊天等．它虽然能一定程度上显示人工智能的力量，但因其并不专注于帮助人解决现实任务问题，其实际使用范围较为狭窄．近年来，随着移动终端的高速发展，面向任务的自然人机对话系统和相关的认知控制理论得到了越来越多的学术和产业界重视，这也是本文讨论的重点

We com-plement recent work by showing the effec-tiveness of simple sequence-to-sequenceneural architectures with a copy mecha-nism. Our model outperforms more com-plex memory-augmented models by 7% inper-response generation and is on par withthe current state-of-the-art on DSTC2, areal-world task-oriented dialogue dataset

Both NLU and NLG are implementedwith template-based models

Symptom ExtractionWe follow the BIO(begin-in-out) schema for symptom identification(Figure 1). Each Chinese character is assigned alabel of ”B”, ”I” or ”O”. Also, each extractedsymptom expression is tagged withTrueorFalseindicating whether the patient suffers from thissymptom or not. In order to improve the anno-tation agreement between annotators, we createtwo guidelines for the self-report and the conver-sational data respectively. Each record is anno-tated by at least two annotators. Any inconsis-tency would be further judged by the third one.The Cohen’s kappa coefficient between two anno-tators are71%and67%for self-reports and con-versations respectively

In this paper, we make a move to builda dialogue system for automatic diagno-sis. We first build a dataset collected froman online medical forum by extractingsymptoms from both patients’ self-reportsand conversational data between patientsand doctors. Then we propose a task-oriented dialogue system framework tomake the diagnosis for patients automat-ically, which can converse with patients tocollect additional symptoms beyond theirself-reports. Experimental results on ourdataset show that additional symptoms ex-tracted from conversation can greatly im-prove the accuracy for disease identifica-tion and our dialogue system is able tocollect these symptoms automatically andmake a better diagnosis

To overcome this issue, weexplore data generation using templates and terminologies and data augmentationapproaches. Namely, we report our experiments using paraphrasing and wordrepresentations learned on a large EHR corpus with Fasttext and ELMo, to learn aNLU model without any available dataset. We evaluate on a NLU task of naturallanguage queries in EHRs divided in slot-filling and intent classification sub-tasks.On the slot-filling task, we obtain a F-score of 0.76 with the ELMo representation;and on the classification task, a mean F-score of 0.71. Our results show that thismethod could be used to develop a baseline system

For the slot filling task, the input is the sentence consisting of a sequence of words, L, and the output is a sequence of slot/concept IDs, S, one for each word. In the statistical SLU systems, the task is often formalized as a pattern recognition problem: Given the word sequence L, the goal of SLU is to find the semantic representation of the slot sequence 푆that has the maximum a posterioriprobability 푃(푆|퐿).

bi-directional Jordan-type network that takes into account both past and future dependencies among slots works best

Open-Domain Neural Dialogue Systems

Open-Domain Neural Dialogue Systems

the dataset used in our experiment hasonly the tags of filled information slots extracted by patternmatching between dialogue log and final order information

the agent model swaps the in-put and output sequences, and it also takes the tag of filledinformation slots as an input which is extracted from dia-logue in previous turns by pattern matching with the orderinformation in ground truth

By directly learningfrom the raw dialogue logs, the network takes the agent ut-Figure 2: The network structure: encoder-decoder structurewith the attention mechanismteranceXa:xa1;xa2;:::;xanas the input sequence and takescorresponding user utteranceYu:yu1;yu2;:::;yumas the tar-get sequence.

In the task-oriented dialogues, a user usually firstly showsthe intention to the agent and then answers the agent’s ques-tions one by one to specify the demand.

we propose a uSer andAgent Model IntegrAtion (SAMIA) framework inspired byan observation that the roles of the user and agent models areasymmetric. Firstly, this SAMIA framework model the usermodel as a Seq2Seq learning problem instead of ranking ordesigning rules. Then the built user model is used as a lever-age to train the agent model by deep reinforcement learning.In the test phase, the output of the agent model is filtered bythe user model to enhance the stability and robustness. Ex-periments on a real-world coffee ordering dataset verify theeffectiveness of the proposed SAMIA framework.