General questions
The only thing necessary to use any of our products is to create an account in MeaningCloud. It takes only a minute, and it's free!
You can also log in with your existing LinkedIn or GitHub account!
A key is generated when you register and you can find it in your personal area.
There are two possibilities:
- The mail you initially gave us had a typo or was incorrect. You can test this by trying to create a new account again; if you get an error, then the email given was ok, and the issue is with the validation mail, which bring us to...
- There was an issue with the validation mail: the first thing to check is if it has been incorrectly sent to your Spam folder. If it hasn't, contact us, and we will solve the issue as soon as possible.
In each product, there's a section in the sidebar that lists the languages it applies to. We currently give full coverage for English, Spanish, French, Italian, Portuguese and Catalan, as well as partial coverage for Danish, Swedish, Norwegian and Finnish.
In the supported languages section you can find a detailed table of the supported languages by API and functionality.
The integrations such as MeaningCloud Add-in for Excel implement these products, so the language support is the same as the one in the table.
There are! In the developers' area there's a section called customization where you can access the different types of user resources that you can define and use with our APIs.
Currently you can define your own version of the following resources:
- Classification models, which you can use with the Text Classification API.
- Deep Categorization Models, which you can use with the Deep Categorization API.
- Dictionaries which can be used in several APIs: Topics Extraction, Sentiment Analysis, Lemmatization, PoS and Parsing and Deep Categorization
- Sentiment Models, which you can use with the Sentiment Analysis API.
Of course! We recommend using our Excel add-in, MeaningCloud Add-in for Excel. It provides the basic functionality of all our APIs, but integrated in Microsoft Excel, allowing you incorporate MeaningCloud easily into your workflow.
If Excel is not your cup of tea, we provide a bunch of other integrations that also enable you to use text analytics without coding.
Sure! We at MeaningCloud always support research and development activities, so if you are working on something, we'd love know what it's about. We'd also really appreciate it if you could share the results of your research project with us once you are finished.
If you were to need elevated access to any of the services, we'd only need a letter of intent describing your planned use of the service and an estimation of requests you need. The letter must be signed by your supervisor at your university.
We also would ask that you mention us in both your publications and on the website of your group/lab. A suggested format would be:
-
- In publications:
MeaningCloud™ (http://www.meaningcloud.com) has been used for Text Analytics purposes in the development/testing/validation of this research/prototype/software.
-
- In the web site of the research group/lab:
This research group/lab uses MeaningCloud™ (http://www.meaningcloud.com) for Text Analytics.
We have reorganized some of the APIs we had available in Textalytics. While most of the features are a available as MeaningCloud APIs or integrations, we have retired some from the public domain and provide only private limited access. These APIs are:
- Speech Recognition and Speaker Diarization API
- User Demographics API
- Semantic Linked Data API
- Media Analysis API
- Semantic Publishing API
Our solutions focus mainly on REST-based web services, but there are additional alternatives for integration in on-premises environments. See more information here. These options are part of our Enterprise plan.
If you are interested in any of them, just write us to sales@meaningcloud.com.
Of course! We provide a free plan that allows you to use all our products for free. The only thing you need to do is to create an account in MeaningCloud and start analyzing!
Aside from the free plan, we provide four different paying plans to cover those scenarios in which the requirements are more demanding. These requirements may be related to volume of text to process, rate limiting, support offered, etc.
In order to buy any of the paying plans, you just need to create an account (if you haven't already), and then click on upgrade, where you will be able to add the billing data and then proceed to subscribe the plan you want.
The plan subscription will start the same day the first payment is made, and it will continue that way until the day a month goes by after the subscription is canceled. In other words, if the first payment is done the 20th of April, and then the subscription is cancelled the 5th of November, then the plan will end on the 20th of November.
The consumption is measured in requests. Different princing plans allow a different number of requests. so, what's a request? This is how we define it:
A request is a HTTP request to any of our services to analyze a text 500 or less words long.So, what happens if the text sent is longer than 500 words? In those cases, the number of requests consumed will be the equivalent of what we would need to analyze that text in chunks of 500 words (that is, requests). For instance, if we are using the excel add-in and in one of the cells we are analyzing we have a text that's 800 words long, then analyzing that cell would count as 2 requests. The 500-word limit applies to all the public APIs except for Language Identification which has a flat rate. Premium APIs have their own limit: for instance, Deep Categorization has a limit of 125 words per request.
The consumption counter is reset the day it makes a month since you purchased your plan, which in the case of the free plan is the day you create your MeaningCloud account.
For users subscribed to a paying plan, there exists the option of restarting the credits early int he subscription page.
Currently payments are carried out through Stripe. Stripe allows you to set recurring subscriptions only by entering your credit/debit details, so each month for as long as the subscription is not canceled, we can charge your credit card the monthly fee without the need to ask you repeatedly for any kind of payment details.
We do not store or access the payment details at any point in the process, they are always managed exclusively by Stripe. Stripe can be used without having to set up an account and as a PCI Level 1 Service Provider offers seamless security in your transactions.
All the details about our data protection policy are explained in the data protection policy section.
There are two types of emails you may receive from us:
- Support emails: which includes mails related to your MeaningCloud account and your use of our services. The ones related to changes in your basic information (for instance, if you reset your password) are not optional, but you can choose not to receive the ones associated to the use of the service (consumption alerts, mostly) in your account.
- Newsletters: where we tell you what we have been up to. They will probably be only a handful a year, and you can always opt out in your account.
This value indicates the number of requests per second you will be able to carry out depending on the plan you are using. In other words, it affects the number of concurrent requests you application can do.
MeaningCloud is designed as a set document processing services that given some text it will produce detailed linguistic and semantic analysis. We provide functionality to analyze from a variety of document formats (html, doc or pdf among others) by extracting relevant texts.
We also provide functionality to analyze a given URL by extracting text, but be aware that the service is not designed for large-scale crawling.
In contrast, if you require to extract meaning from huge public document collections, the web or social networks streams we can provide consultancy on using MeaningCloud products with related solutions. Contact us if you are interested.
Technical questions
Depending on the environment in which you are installing the add-in, which other Microsoft applications you have installed, etc. there may be issues with the installation process.
These are some of the things you can check:
- All Microsoft programs are closed when you carry out the installation.
- You have the necessary rights to install Microsoft Office customizations.
- There's not another version of the add-in installed (in which case, you should uninstall it before installing the new one).
2. The Excel add-in seems to have been installed successfully but I cannot find it when I open Excel
Sometimes add-ins are disabled and they stop showing in Excel's toolbar.
To check if this is the case, you have to go to Excel's options (usually a button, in the Home or Office button, depending on the version). In it, there's an Add-ins tab where you can check two different list, just by selecting them next to Manage at the bottom of the menu:
- COM Add-ins: if the add in is listed here, you have to make sure it's checked.
- Disabled elements: if the add-in is in this list, you have to enable it.
When the "Cannot connect to service" message error appears, it is usually related to connectivity problems. There are two main reasons why this happens:
- The machine where the add-in is installed is not connected to the internet, which is a requirement for it to work.
- The internet connection is done through a proxy server, and it has not been configured in the add-in. To fix this you just have to configure the connection settings of the add-in with the proxy server details.
Microsoft sells certificates to identify applications as trusted developments. If you don't buy them, then it does raise a message in the installation informing you that Windows does not know/trust the developer. In most cases this should not be an problem when it comes to installing it the add-in.
The only instance where you may have issues is if the policies in your machine restrict installations from non-trusted developers, in which case you will probably will have to contact your IT team to see how to proceed.
MeaningCloud defines two different types of main semantic units, concepts and entities.
Entities are defined as real-world objects that you can identify with a proper name, for instance 'Barack Obama' or 'Spain'. On the other hand, concepts are terms that help to describe the content of a text and that among other things may serve to describe classes of entities, like 'president' or 'country'.
Our definition of concepts goes beyond plain nouns and we also include other terms, like adjectives, that provide a semantic interpretation of the content of a text.
Both concepts and entities are linked to nodes in our ontology, 'Spain' is a 'Top>Location>GeoPoliticalEntity>Country' as well as 'country' as a term describes 'Top>Location>GeoPoliticalEntity>Country' in English. Likewise, 'país' does the same in Spanish.
In the ontology and knowledge representation literature entities are represented as instances while Natural Language Processing literature talks about the problem of Named Entity Recognition. On the other hand, concepts are the terms that help to describe classes of instances. However, there are grey areas for general knowledge representation theory as well as the criteria may differ for your particular application. Check out for particular examples and be aware that our criteria leans towards the NLP side and ground in our previous experience on practical applications.
Mentions of entities can be polysemic the same way as common words are. For instance, 'London' can refer to both the city and to the surname of "The Jungle's Book" author "Jack London" among others. MeaningCloud analyzes the context in which a particular mention of an entity appears and tries to disambiguate to the correct reference, the city or the person in the case of 'London'. MeaningCloud uses an internal knowledge base which identifies each entity with an unique id as well as classifies as part of our ontology of entity types.
Besides we provide links to some of the most common public knowledge bases like Wikipedia, Freebase or Dbpedia. Some of this knowledge bases use Linked Data standards to publish related information. By using MeaningCloud you can crosslink to these public knowledge bases and make your content truly semantic.
The disambiguation process is carried out so through semantic, grammatical and morphological cues we are able to extract which meaning the entity has in a given context.
The entities detected by all our APIs follow the our own ontology, more specifically, the entity branch, which can be easily explored in this chart.
We have a support section to manage any kind of feedback from our users, including bugs. In the form, there's a field called "Type", where one of the possible values is "Bug report".
To report a bug, you just have to contact us through that form selecting the "Bug report" value, and we will proceed to study and correct it in future versions.
Automated language analysis is definitely a complex matter. We aim to provide state-of-the-art tools for linguistic and semantic analysis but errors and limitations still exists. Send us those MeaningCloud analysis that do not make sense to you or are just plain weird.
We have a support section to manage any kind of feedback from our users, including bugs. In the form, there's a field called "Type", where one of the possible values is "Bug report". Once you select that value and the service in which you have found the annotation, you will be able to specify that the reason is related to an "Incorrect item", and give more details about it.
Through these details we will be able to analyze why the annotation is incorrect and fix it in future releases.
This error is related to the license key you are using. The first thing you should do is to make sure you are using the correct license key, checking there are no additional characters where you are using it (in one of our integrations or in your own code).
If the error persists, write us and we will check it out as soon as possible.
Every API has a limit on the number of words that can be analyzed in a single request, much like the size limit operations such as POST have. In our case, we use words as the way to measure this limit, so there's less of a dependency on the format used to process the text.
4. I get an error saying that I've exceeded the maximum number of requests per second. What do I do?
This limits the number of request you can carry out concurrently. If you are sending the requests sequentially and you are getting this error, a solution is to establish a throttle (pause) between calls, ensuring they proceed below your plan rate limits.
If you require to proceed with high speed and high volume whatsoever, check our plans and pricing, or write us with your requirements to work out a solution.
