HappyDB is a corpus of 100,000+ crowd-sourced happy moments. The goal of the corpus is to advance the state of the art of understanding the causes of happiness that can be gleaned from text. Please see also our web site
We conducted a large scale collection of happy moments over 3 months on Amazon Mechanical Turk (MTurk.) For every task, we asked the MTurk workers to describe 3 happy moments in the past 24 hours (or past 3 months.)
Here are the instructions of the data collection task.
What made you happy today? Reflect on the past {24 hours|3 months}, and recall three actual events
that happened to you that made you happy. Write down your happy moment in a complete sentence.
Write three such moments.
Examples of happy moments we are NOT looking for (e.g events in distant past, partial sentence):
- The day I married my spouse
- My Dog
Basic statistics of HappyDB are shown in the table.
| Collection period | 3/28/2017 - 6/16/2017 |
|---|---|
| # happy moments | 100,922 |
| # distinct users | 10,843 |
| # distinct words | 38,188 |
| Avg. # happy moments / user | 9.31 |
| Avg. # words / happy moment | 19.66 |
You can download the dataset by using git command or simply downloading the file from the repository.
$ git clone <git-repository-path>
After you clone or download the repository, you will see the following file structure.
happydb
└── data
├── cleaned_hm.csv
├── demographic.csv
├── original_hm.csv
├── senselabel.csv
├── topic_dict
│ ├── entertainment-dict.csv
│ ├── exercise-dict.csv
│ ├── family-dict.csv
│ ├── food-dict.csv
│ ├── people-dict.csv
│ ├── pets-dict.csv
│ ├── school-dict.csv
│ ├── shopping-dict.csv
│ └── work-dict.csv
└── vad.csv
HappyDB consists of a set of CSV files. Here are schema descriptions of the files.
cleaned_hm.csv contains cleaned-up happy moments and some additional information in addition to original happy moments.
- hmid (int): Happy moment ID
- wid (int): Worker ID
- reflection_period (str): Reflection period used in the instructions provided to the worker (3m or 24h)
- original_hm (str): Original happy moment
- cleaned_hm (str): Cleaned happy moment
- modified (bool): If True,
original_hmis "cleaned up" to generatecleaned_hm(True or False) - predicted_category (str): Happiness category label predicted by our classifier (7 categories. Please see the reference for details)
- ground_truth_category (str): Ground truth category label. The value is
NaNif the ground truth label is missing for the happy moment - num_sentence (int): Number of sentences in the happy moment
original_hm.csv contains unfiltered version of happy moments.
- hmid (int): Happy moment ID
- wid (int): Worker ID
- hm (str): Original happy moment
- reflection_period (str): Reflection period used in the instructions provided to the worker (3m or 24h)
demographic.csv contains demographic information of the workers who contributed to the happy moment collection.
- wid (int): Worker ID
- age (float): Age
- country (str): Country of residence (follows the ISO 3166 Country Code)
- gender (str): {Male (m), Female (f), Other (o)}
- marital (str): Marital status {single, married, divorced, separated, or widowed}
- parenthood (str): Parenthood status {yes (y) or no (n)}
senselabel.csv contains multi-word expression and supsersense tags on "cleaned" happy moments. Thus, the number of rows are exactly same as that of cleaned_hm.csv.
- hmid (int): Happy moment ID
- tokenOffset (int): Position index of a token
- word (str): Token in the original form
- lowercaseLemma (str): Lemmatized token in lowercase
- POS (str): Part-of-Speech tag
- MWE (str): Multi-word expression (MWE) tag in the extended IOB style (See [REF] for further information)
- offsetParent (int): The beginning position of a multi-word expression
- supersenseLabel (str): Supersense classes defined in the WordNet (19 verb and 25 noun classes. See [REF] for further information.)
Here is an example of annotated happy moments. A sentence is tokenized into words. You may see that "got" in the 2nd position is lemmatized as "get" in the lowercaseLemma column. The supersenseLabel column shows the supersense labels of each word or multi-word expression.
hmid tokenOffset word lowercaseLemma POS MWE offsetParent supersenseLabel
0 70027 1 I i PRON O 0 NaN
1 70027 2 got get VERB B 0 v.motion
2 70027 3 my my PRON o 0 NaN
3 70027 4 car car NOUN o 0 n.artifact
4 70027 5 waxed waxed NOUN I 2 NaN
5 70027 6 and and CONJ O 0 NaN
6 70027 7 polished polish VERB O 0 v.motion
7 70027 8 . . PUNCT O 0 NaN
happydb/data/topic_dict contains the topic (e.g., entertainment, exercise etc.) dictionaries that are manually prepared for analyzing topics of happy moments. Each file contains keywords that belong to the topic.
All files (except for *-dict.scv files) follow the CSV-format. Each programming language should have some library that has CSV loading function. In Python, pandas is a common library for handling data.
Here are some examples looking at data with pandas library.
>>> import pandas as pd
>>> df = pd.read_csv("happydb/data/cleaned_hm.csv")
>>> df["cleaned_hm"].head()
0 I went on a successful date with someone I fel...
1 I was happy when my son got 90% marks in his e...
2 I went to the gym this morning and did yoga.
3 We had a serious talk with some friends of our...
4 I went with grandchildren to butterfly display...
Name: cleaned_hm, dtype: object
>>> df["predicted_cattegor"”].value_counts
affection 34206
achievement 34044
enjoy_the_moment 11202
bonding 10729
leisure 7505
nature 1846
exercise 1206
Name: predicted_category, dtype: int64We plan to release sample scripts and iPython (Jupyter) notebooks for the HappyDB soon!
Please cite the following publication if you use the dataset in your work.
Akari Asai, Sara Evensen, Behzad Golshan, Alon Halevy, Vivian Li, Andrei Lopatenko,
Daniela Stepanov, Yoshihiko Suhara, Wang-Chiew Tan, Yinzhan Xu,
``HappyDB: A Corpus of 100,000 Crowdsourced Happy Moments'', LREC '18, May 2018. (to appear)
Please ask us questions at happydb@recruit.ai.