You can type any text, into the input text box:

Note that you also have alternative ways of providing text, such as the microphone (only Windows) or loading a text file :

In case of the file loading, you have to choose if split the content of the file, assuming the sentences are line by line or split them a natural way where the dot is the indicator of the end of sentence.

And when the "PROCESS" button is clicked, then the provided text is cleaned, splitted into valid sentences, analyzed and stored into the memory their identified relations:

Unless there is any problem with the input:

The valid sentences will be listed in the right box, and it has associated an icon, that could be an green icon indicating the sentence has been correctly processed or an red icon that means that something wrong was happened (E.g.: the NLP server is not listening at that concrete moment)

In case of any error, it appears a button for reprocessing those failed sentences.

When you select any of the processed sentences, the GUI will show a description of the processing results.

• That are their syntax

      parsed = the words of the sentences after being cleaned

      lemmas = basic form of the word

      types = POS (Part of the Speech) = the type of word

      Entities = NER (Named Entity Recognized) = if the word has been identified as Person, Location, Organization, Number ...

  
  

• The constituent tree, or what it's the same the grammar, with highlighted those words that has been identified as subject, verb or object.

  
  

• And the extracted relations in ILC format.

  
  

In case you select one that has failed (shows the red icon), it will show the error ocurred during its processing.

Finally, you can query the already processed sentences:

Choosing between latest provided or searching for a certain word(s) or combinations of words, using the symbol | that means OR, or the symbol & that means AND. The number of maximum sentences returned is limited by the indicated number. [@d]

Take in mind, the system stores the sentences after being correctly processed, so each word of the original sentences is replaced by their lemma.

Then you should made the searchs in lowercase, avoid the plurals, the numbers are replaced by integers, each word that does not contains letters or numbers is removed, and capital letters (only the first letter) are permitted for propers.E.g.: The cats of Anne MARIE are 10,9 % mammals → the cat of Anne be 11 mammal.

For maintenance reasons, in case any processed sentence suffered a syntax or grammar error, this is stored in the database for a posterior analysis.

Note: if you choose option "clear", all the sentences will be removed from this list, and can't be recovered.

This section mainly manages [Tp], [Tl] and [Tf] orders

The first form correspond with the affirmative question type.

Where you can write the concepts and select the valid options (verb keys).

The question form changes based in the key value; For example, if the selected key is HAVE then appears the option to input numbers.

Each time a concept input box is modified, it appears a list with the matching concepts the system has (autocomplete function).

When the form is populated the question button ("?") is shown, and can be clicked (also pressing the return key) to perform the question.

Note the indicators that appears below each concept input box when the focus is lost:

    green tick = the word(s) are valid and exists into the lexicon

    yellow warning = the input doesn't match with any existing concept

When the question is asked, the response is shown.That it could be an error

Or the answer (Yes, No, Unknown).

Do not forget the "Deep Search" options (see at the top-right of the tab), since affects the answer.

Moreover, in this kind of question, You can confirm if the answer is right or not.

The "True/False"confirmation buttons only can be pressed once.

If you want to check if the system learned this, you can ask again the same question and see, in case of correcting, if the answer changed (take in mind, the question button remains disabled until the question form is modified). Or you can validate the relation, exploring the content of the concept.

As the relation affects directly the concept frame (nor the attribute or parent that matched the conditions). In case any filter is activated (see at the top-right of the tab), a warning icon is shown indicating that the response of the question could be determined by those limitations.

If the search returns a non "Unknown" response, and the asked characteristic is not found directly in the concept frame. Unless any filter is active, it appears the button "Show Path" ; if you press it then you can visualize the route traveled until reaching the concept that matched the question.

This section manages [?a] and [?c] orders

The second form allows group questioning.

The interaction is the same as above, you can select the question key, input the concepts (being helped by the autocomplete and the concept flags) and perform the question. Note that you can change from one input box to another using the "tab" key, and the grey button is for clean and reset the question form.

Remember that the subject (first input concept box) in this type of question is optional.

Unless there is an input issue or no concepts has fitted the asked conditions, a list of concepts that satisfied the question will be shown.

You can access directly to their content by clicking on them.

For security reasons the results are limited up to 20 concepts.

This section manages [?s] orders

The next form allows numbered attribute questioning.

In this case you have to indicate both concepts and receive the answer,

that can be a list of the associated numbers to the indicated attribute relation; "Any" if the relation exists but no numbers are indicated;

"None" if doesn't exist that relation or an issue description in case of validation error

This section manages [?n] orders

This group box allows to question about interaction groups.

In this case you have to indicate the kind of question type:

What DO(es) {not} CONCEPT ACTION ?

or What CAN {not} ACTION CONCEPT ?

that will answer the question with the list of interactions that fits the question (limited up to 20 concepts).

This section manages [?i] orders

The next group box allows to question about affirmative interactions.

In this case you have to indicate the concepts, the action verb and the receiver concept,

that will answer the question with a "Yes" or "No" and gives the opportunity of confirm the fact in the same way as affirmative kind question does.

If the question is satisfied but the relation is not found directly in the concept frame. Unless any filter is active, it appears the button "Show Path" ; if you press it then you can visualize the route traveled until reaching the concept that matched the question.

This section manages [?t] orders

And the last one it's a free input box where you can type any kind of question.

Where you can write any question, or even dictating via voice:

This feature is only available for Windows OS, it requires the Windows Speech API and a correctly installed microphone.

And after click on the "?" button (or pressing the enter key), and it will show the results.

That it could be an error description if there is a problem, such as, the question has not a valid and accepted format.

or showing the answer (and it populates the question values) into the correspondent box, based into the kind of asked question:

In case of being a correct multiple condition question, it moves to the correspondent guess tab

Note that the search parameters of the guess tab will be modified. Setting "strict" as ticked and "deep search" with the same values as this tab has.

This section manages [?f] orders

You can select the valid options, write the concepts (obviously being helped by the autocomplete and icons), adding new conditions (pressing the "+" button), removing conditions ("-" button), clicking on the "reset" button, if you want to clean the form . . .

You also have the option of typing the question in raw text:

When there was inputted a valid multiple condition question (in the form or in the free box question text), the "?" button to perform the question is shown.

And if you click on them or press the enter key then the answer of the question will be displayed. But take in mind the the free box has preference, so everything you had typed in the form will be replaced by the typed question in the free box. Since the form values are populated with the typed question.

Don't worry about leaving empty condition lines (considered when their concept is empty) or not consecutive, as the application it removes automatically when the question is performed. Anyway, only for the GUI available size, there is a limitation of 10 condition as maximum.

Since it affects a lot the results and answers, when the question is asked, check the search conditions, both the strict and deep search checkboxes.

And do not forget the tendency and source filters.

When a question is asked you can receive the following responses:

• And error description if the question is not correctly typed, there has a non existent concept, is not an accepted type, or even if it's a valid question but not a multiple condition group question:

  







• In case the question is well formulated but no concepts had fit into the conditions:

  







• Or the list of concepts that has a hit ratio higher than zero over the requirements of the question:

  

  
  

  
  

  There are also two result filters: Threshold and Limit (though the results are always limited to a maximum of 500 concepts)

  

  
  

  
  

  In the case of strict search, as each concept has 100% of hit ratio, the "threshold" and the " % Hit "it doesn't have sense, so are disabled:

  

  
  

  Remind that you can access directly to their content by clicking on them.

This section manages [?g] and [?f] orders

When you type a sentence it will draw the sentence highlighting in bold the semantics words,

placing in boxes the identified subjects, verbs and objects, and creating arrows between the extracted relations and characteristics.

It will show the correspondent error in case an invalid sentence or a sentence without relations has been inserted.

On valid sentences, it will list the identified internal language codes of the relations (it matches with the arrows drew above the sentence)

and the rest of identified characteristics (the arrows printed below the sentence).

The items of the lists will have associated a "green tick" indicating the relation exists, a "yellow cross" means the relation exists but with different tendency as the sentence indicates; and a "red cross" means the relation does not exists.

If you put the mouse over the icon, a little explanation is shown.

If the item is a "red or yellow cross", you can click in the icon to append that relation into the system ("force to true" option).

If it's not a "red cross", then clicking the icon it will graph the relation path ("see relation path" option).

Finally it provides the sense of the identified subjects.

This means that if a concept has more than one parent, depending of the characteristics extracted, it decides what is the correct meaning for the context of the inputted sentence.

E.g.: "that mouse is broken" → the sense of mouse in this case is a PC device. On the other hand, "the mouse ate all the cheese" → the sense is mammal or animal in this context.

A "yellow cross" means the sense couldn't be identified, when is "green tick" will show the associated concepts with the confidence the system has about it.

This section mainly manages [Ta], [Tw], [Ty] and [Ti] orders

You can append new concepts from the WORD section:

The content of the input boxes is validated, when both boxes are right (green check appears) the button of "save" goes to enabled.

A word/lemma is valid when only contains English letters (a-z), it could be splitted by an dash symbol (e.g.: warm-blooded) or by an white space (multiple word concept such as: basketball team). The word is allowed to be propercase (the first letter in uppercase; e.g. Bird). Never in the case of the lemma.

When the "save" button is pressed the results of the processing is shown; "OK" if everything is fine; "KO" otherwise.

Also, there is an specific guided section for the relations between the concepts:

The "save" button appears when both concepts are valid, in other words, it exists in the dictionary/lexicon.

An optional extra could be inputted in case of:

- attribute(HAVE): comma separated list of numbers or ranges. E.g.: 1,4,5-15,22

- skill(CAN): a comma separated list of interactions with the format [existent concept*tendency]. E.g.: fly*1,swim*-2

A green check is shown under the extra box when the valid value is inputted.

Obviously, you have the option of typing the entire internal language code:

In case the check box is active the system also will create any word and relation needed to append correctly the indicated knowledge.

Or processing a file:

This section mainly manages [Ti] and [Tc] orders