Predict
Model Predictions
To make predictions, you can simply use pretrained or custom models by simply utilizing as a table under the model schema. Superinsight support the following models.
| Task | Description | Model Name |
|---|---|---|
| Text Classification | Use to assign label or class to text, labels such as sentiment or product category | model.text_classification |
| Question Answering | Use to find answers in a large amount of text | model.question_answering |
| Recommender | Use to recommend items to user base on collaborative filtering | model.recommender |
| Summarization | Use to summarize a large amount of text to a short description | model.summarization |
| Text Generation | Use to generate text from a preceding example | model.text_generation |
| Translation | Use to translate one language to another, for example English to French | model.translation |
Text Classification
Text classification can be use to generate labels from text. For example, let's say you want to determine if movie overview are consider as are comedy, drama or others, and you run the following query
SELECT mldb.movie._id, predictions.*
FROM mldb.movie
JOIN model.text_classification
ON model.text_classification.inputs = mldb.movie.overview
WHERE mldb.movie._id <= 3
AND model.text_classification.labels = 'comedy,drama,others'
| _id | inputs | score_comedy | score_drama | score_others |
|---|---|---|---|---|
| 1 | Two imprisoned men bond over a number of years, finding solace and eventual redemption through acts of common decency. | 0.0624 | 0.7388 | 0.1987 |
| 2 | An organized crime dynasty's aging patriarch transfers control of his clandestine empire to his reluctant son. | 0.0272 | 0.7299 | 0.24282 |
| 3 | When the menace known as the Joker wreaks havoc and chaos on the people of Gotham, Batman must accept one of the greatest psychological and physical tests of his ability to fight injustice. | 0.0120 | 0.8850 | 0.10289 |
Question Answering
Question answering can be use to find answers from text. For example, let's say you have the description of a movie and you need to find the main character, you can use question answering model to determine that for you.
SELECT mldb.movie._id, predictions.inputs, predictions.question, predictions.answer, predictions.score
FROM mldb.movie
JOIN model.question_answering
ON model.question_answering.inputs = mldb.movie.overview
WHERE mldb.movie._id < 20
AND predictions.score > 0.2
AND model.question_answering.question = 'Name of character'
ORDER BY predictions.score DESC
| _id | inputs | question | answer | score |
|---|---|---|---|---|
| 11 | A meek Hobbit from the Shire and eight companions set out on a journey to destroy the powerful One Ring and save Middle-earth from the Dark Lord Sauron. | Name of character | Hobbit | 0.6807 |
| 8 | In German-occupied Poland during World War II, industrialist Oskar Schindler gradually becomes concerned for his Jewish workforce after witnessing their persecution by the Nazis. | Name of character | Oskar Schindler | 0.2687 |
Recommender
Recommender can be use to predict items that a user will be interested in base on other users with similar interests.
Prerequisite
There is no default model for recommender because recommender model can only be effective after being trained from custom data, to create a custom recommender model, follow instructions from training recommender model for more information.
The following query demonstrate how to recommend movies to a user by joining the table mldb.movie_rating with a custom recommender model {my_model_id}.
Predictions for this model requires the parameters user_id and item_id, and we can pass them into the model by specifing the table column mldb.movie_rating.user_id to be map as model.recommender.user_id and the table column mldb.movie_rating.movie_id to be map as model.recommender.item_id.
To specifiy how many items to recommend, include the condition model.recommender.max_size = 5. The default max_size is 3.
To specifiy the user to predict recommendations, include the condition mldb.movie_rating.user_id = 1. To make recommendations for all users, you can skip this condition.
Since we want to see the top recommendations, we can use predictions.score DESC to sort the recommendations by the highest score.
SELECT predictions.user_id, predictions.item_id, predictions.score, mldb.movie_rating.movie_id
FROM mldb.movie_rating
JOIN model.recommender ON model.recommender.user_id = mldb.movie_rating.user_id
WHERE model.recommender.item_id = mldb.movie_rating.movie_id
AND model.recommender.model_id = '{my_model_id}'
AND model.recommender.max_size = 5
AND mldb.movie_rating.user_id = 1
ORDER BY predictions.score DESC
| user_id | item_id | score | movie_id |
|---|---|---|---|
| 1 | 2204 | 0.0074 | 2204 |
| 1 | 1674 | 0.0071 | 1674 |
| 1 | 91529 | 0.0064 | 91529 |
| 1 | 6932 | 0.0061 | 6932 |
| 1 | 158528 | 0.0059 | 158528 |
Summarization
Summarization can be use to summarize long text to a shorter summary. For example, let's say you need to summary a movie into a short summary, you can use the summarization model to do that for you.
SELECT mldb.movie._id, mldb.movie.title,mldb.movie.overview, predictions.summary
FROM mldb.movie
JOIN model.summarization
ON model.summarization.inputs = mldb.movie.overview
WHERE mldb.movie._id = 12
| _id | title | overview | summary |
|---|---|---|---|
| 12 | Forrest Gump | The presidencies of Kennedy and Johnson, the events of Vietnam, Watergate and other historical events unfold through the perspective of an Alabama man with an IQ of 75, whose only desire is to be reunited with his childhood sweetheart. | An Alabama man with an IQ of 75, whose only desire is to be reunited with his childhood sweetheart. |
Text Generation
Text generation model can be use to create text to get insights from existing data. For example, if we know that a person likes certain movies, we can use the overview of the movies to generate text to recommend related movies. There is a big difference between using text generation to generate recommendations versus using a recommender. Recommender model uses custom data to train a custom model and the model has no common knowledge about the outside world. Text generation is a pretrained model that already has knowledge about the world so it doesn't need custom data to make predictions.
A thief who steals corporate secrets through the use of dream-sharing technology is given the inverse task of planting an idea into the mind of a C.E.O.
Q: If the person likes the above content, list other movies this person will like in comma delimeter format?
A: The Matrix, The Truman Show, The Prestige, Inception, The Dark Knight
SELECT mldb.movie.overview, predictions.prompt, predictions.output
FROM mldb.movie
JOIN model.text_generation
ON model.text_generation.inputs = mldb.movie.overview
WHERE mldb.movie._id = 9
AND model.text_generation.prompt = '\nQ: If the person likes the above content, list other movies this person will like in comma delimeter format?\nA:'
AND model.text_generation.stop_word = '\\n'
| overview | prompt | output |
|---|---|---|
| A thief who steals corporate secrets through the use of dream-sharing technology is given the inverse task of planting an idea into the mind of a C.E.O. | Q: If the person likes the above content, list other movies this person will like in comma delimeter format? A: | The Matrix, The Truman Show, The Prestige, Inception, The Dark Knight |
Translation
Translation model can be use to translate text from one language to another.
See the Supported Language Code section below for all supported languages.
SELECT mldb.movie._id, mldb.movie.title, predictions.inputs, predictions.output
FROM mldb.movie
JOIN model.translation
ON model.translation.inputs = mldb.movie.overview
WHERE mldb.movie._id = 12
AND model.translation.source_language = ['en_XX']
AND model.translation.target_language = ['fr_XX']
| _id | title | inputs | output |
|---|---|---|---|
| 8 | Forrest Gump | The presidencies of Kennedy and Johnson, the events of Vietnam, Watergate and other historical events unfold through the perspective of an Alabama man with an IQ of 75, whose only desire is to be reunited with his childhood sweetheart. | Les présidences de Kennedy et Johnson, les événements du Vietnam, Watergate et d'autres événements historiques se déroulent dans la perspective d'un homme d'Alabama ayant un IQ de 75 ans, dont le seul désir est de se réunir avec son ami de l'enfance. |
Supported Language Code
Arabic (ar_AR), Czech (cs_CZ), German (de_DE), English (en_XX), Spanish (es_XX), Estonian (et_EE), Finnish (fi_FI), French (fr_XX), Gujarati (gu_IN), Hindi (hi_IN), Italian (it_IT), Japanese (ja_XX), Kazakh (kk_KZ), Korean (ko_KR), Lithuanian (lt_LT), Latvian (lv_LV), Burmese (my_MM), Nepali (ne_NP), Dutch (nl_XX), Romanian (ro_RO), Russian (ru_RU), Sinhala (si_LK), Turkish (tr_TR), Vietnamese (vi_VN), Chinese (zh_CN), Afrikaans (af_ZA), Azerbaijani (az_AZ), Bengali (bn_IN), Persian (fa_IR), Hebrew (he_IL), Croatian (hr_HR), Indonesian (id_ID), Georgian (ka_GE), Khmer (km_KH), Macedonian (mk_MK), Malayalam (ml_IN), Mongolian (mn_MN), Marathi (mr_IN), Polish (pl_PL), Pashto (ps_AF), Portuguese (pt_XX), Swedish (sv_SE), Swahili (sw_KE), Tamil (ta_IN), Telugu (te_IN), Thai (th_TH), Tagalog (tl_XX), Ukrainian (uk_UA), Urdu (ur_PK), Xhosa (xh_ZA), Galician (gl_ES), Slovene (sl_SI)